Hello,
This is how many times I have been trying "The Fix" and would appreciate anyone that might know someone that was unable to set Stacksize for Linux Boxes?
Presently am crunching "Simap" and would really appreciate any advice anyone could offer?
Sincerely
Doug Worrall
Docking@Home
Powered by
Results for user
Next 20
Result ID
click for details Work unit ID
click for details Sent Time reported
or deadline
explain Server state
explain Outcome
explain Client state
explain CPU time (sec) claimed credit granted credit
22462 6305 25 Sep 2006 23:53:21 UTC 26 Sep 2006 1:00:42 UTC Over Client error Compute error 329.12 0.43 ---
22458 6303 26 Sep 2006 1:00:42 UTC 26 Sep 2006 11:43:17 UTC Over Client error Done 337.12 0.44 ---
22421 6291 25 Sep 2006 21:51:59 UTC 25 Sep 2006 23:53:21 UTC Over Client error Compute error 334.30 0.44 ---
22307 6253 26 Sep 2006 20:11:27 UTC 26 Sep 2006 22:23:14 UTC Over Client error Done 323.29 0.42 ---
21987 3570 25 Sep 2006 20:45:01 UTC 25 Sep 2006 21:51:59 UTC Over Client error Compute error 331.96 0.43 ---
21926 3554 24 Sep 2006 13:08:00 UTC 24 Sep 2006 16:09:33 UTC Over Client error Compute error 329.79 0.43 ---
21894 3397 24 Sep 2006 4:55:01 UTC 24 Sep 2006 5:01:11 UTC Over Client error Compute error 321.28 0.42 ---
21857 3710 23 Sep 2006 21:39:16 UTC 23 Sep 2006 22:58:06 UTC Over Client error Compute error 325.42 0.42 ---
21851 2580 23 Sep 2006 20:31:57 UTC 23 Sep 2006 21:39:16 UTC Over Client error Compute error 330.05 0.43 ---
21845 3587 23 Sep 2006 19:00:30 UTC 23 Sep 2006 19:17:27 UTC Over Client error Done 333.46 0.44 ---
21745 5896 22 Sep 2006 11:40:55 UTC 22 Sep 2006 11:47:44 UTC Over Client error Compute error 325.64 0.42 ---
21351 5901 21 Sep 2006 19:40:25 UTC 21 Sep 2006 19:47:01 UTC Over Client error Compute error 336.14 0.42 ---
21345 3708 22 Sep 2006 11:47:45 UTC 22 Sep 2006 12:54:03 UTC Over Client error Compute error 335.97 0.44 ---
21165 5874 21 Sep 2006 18:03:23 UTC 21 Sep 2006 18:09:39 UTC Over Client error Compute error 325.85 0.41 ---
21106 5904 21 Sep 2006 18:29:44 UTC 21 Sep 2006 18:36:01 UTC Over Client error Compute error 329.48 0.42 ---
20930 5867 20 Sep 2006 16:15:38 UTC 20 Sep 2006 16:26:52 UTC Over Client error Compute error 328.23 0.41 ---
20913 5928 20 Sep 2006 16:26:52 UTC 20 Sep 2006 16:33:13 UTC Over Client error Compute error 335.17 0.42 ---
20784 3588 21 Sep 2006 18:44:43 UTC 21 Sep 2006 18:51:16 UTC Over Client error Compute error 323.51 0.41 ---
20777 3788 21 Sep 2006 15:37:32 UTC 21 Sep 2006 15:43:58 UTC Over Client error Compute error 334.27 0.42 ---
20736 3675 21 Sep 2006 15:20:37 UTC 21 Sep 2006 15:27:35 UTC Over Client error Compute error 327.94 0.41 ---
Doug,
I was having the same problems and I could not get the 'ulimit' thing to do much for me.
I posted a question in "Questions & Answers", Linux, Distro Questionaire and Andre gave back this information:---
Find a file called run_manager in your BOINC installation directory and add the line 'ulimit -s unlimited' (without the quotes) as the first line in that file using your favorite editor. Now stop your running boinc manager, open a terminal, type 'cd <boinc installation directory> followed by ./run_manager. You might also be able to use your GUI file manager to go to that directory and double-click on the run_manager file.
Following these instructions I was able to get things working. I did not use the terminal mode to run the Boinc programme as I used a GUI in Fedora Core 3.
if you have trouble following this post go to the Qustions & Answers where you reported your problems with Linux uder the Post "Distro Questionaire" and go down about 3/4 of the way till you get to my question asking how to use "ulimit" and then follow Andre's response.
Doug,
I was having the same problems and I could not get the 'ulimit' thing to do much for me.
I posted a question in "Questions & Answers", Linux, Distro Questionaire and Andre gave back this information:---
Find a file called run_manager in your BOINC installation directory and add the line 'ulimit -s unlimited' (without the quotes) as the first line in that file using your favorite editor. Now stop your running boinc manager, open a terminal, type 'cd <boinc installation directory> followed by ./run_manager. You might also be able to use your GUI file manager to go to that directory and double-click on the run_manager file.
Following these instructions I was able to get things working. I did not use the terminal mode to run the Boinc programme as I used a GUI in Fedora Core 3.
if you have trouble following this post go to the Qustions & Answers where you reported your problems with Linux uder the Post "Distro Questionaire" and go down about 3/4 of the way till you get to my question asking how to use "ulimit" and then follow Andre's response.
Hello Conan,
Got this mesage after I typed "ulimit -s unlimited" in {run_manager} then from console got this message
Password:
[root@localhost ~]# cd<boinc installation directory>
-bash: syntax error near unexpected token `newline'
[root@localhost ~]# ./run_manager
-bash: ./run_manager: No such file or directory
[root@localhost ~]#
I assume seeing there are 2 run_manager, but one is a text file so put the ulimit in there with Kedit,in the first line ? have done that before, and then before also tried
the same in Console, are the greater then supposed to be there?Have been at this for a week without success, there must be something silly I am doing wrong, cause I am the only one that cannot make this work, So will keep at her,Thanks
Conan.Do you "enter" the info in console after typing in <boinc installation directory>, then ./run_manager? Then enter?
Hello Doug,
As I am not too good with the text syntax method of running programmes, I just used the GUI fedora interface.
I went to the Boinc folder and open it.
I then located the "run_manager" file and opened it in my text editor by right clicking on the file.
Added the line "ulimit -s unlimited" (without the exclamation marks), as the first line of the file and then saved the file.
Stopped BOINC Manager programme.
I then restarted the Boinc Manager programme again mamually from inside the folder (as I have not yet worked out how to get the Linux system to reboot when power goes off, and how to automatically get it past the login screen and get it to start a specific programme, without me being there, but I will one day).
Things started to work properly after this.
I am not sure how to run the programme from the terminal screen.
hope this helps.
____________
Looks like the directory path has some strange characters... Or that you are typing <boinc installation directory> literally instead of replacing with the directory where your BOINC is installed ;)
OK, a few things to try to narrow down the problem. In cases where the sentence ends in a path or filename, I've put a space before the period that terminates the sentence, so as not to make you think it's part of the path or filename. Note that boincmgr is the actual program that run_manager does setup for and then runs.
1. Where do you have the boinc client installed? On my systems, it's something like
/home/BOINC
or
/BOINC
. That's where you should find run_manager and it should be a text file. I only have one file named run_manager on my system. That second copy you say you have is kind of strange. Could you type the following command at a console while running as root and post the output.
locate run_manager
2. If you're using a desktop icon to start the BOINC manager, right click on the icon, click on properties, and see what command it executes. On my FC3 system, it's
/home/BOINC/run_manager
. That's the file you should edit. You should stop BOINC, edit the file, and restart BOINC by clicking on the icon.
3. Are you running the BOINC client as root or have you setup a separate user for it? If you're letting boincmgr start the BOINC client then, using whatever user you're logged in as when you run the boincmgr, type the following command and post the output. This will show the default limits for that user.
ulimit -a
4. I looked at the computer in your profile and it says you're running Linux 2.6.16.27.tex1.lve . Do you know if you have selinux enabled?
That should be enough to get started. I know there's some security limits on Linux 2.6.x about changing values with ulimit. I'll have to find where to change that again. There's a soft and a hard limit and then there's a file somewhere that let's you override that for certain users. I ran across some info on that the other day, but will have to find it again.
Regards,
-- David
EDIT: Left something out of one question and deleted quoted text since it was large. See the message this is in response to if you need to reference it.
OK, a few things to try to narrow down the problem. In cases where the sentence ends in a path or filename, I've put a space before the period that terminates the sentence, so as not to make you think it's part of the path or filename. Note that boincmgr is the actual program that run_manager does setup for and then runs.
1. Where do you have the boinc client installed? On my systems, it's something like
/home/BOINC
or
/BOINC
. That's where you should find run_manager and it should be a text file. I only have one file named run_manager on my system. That second copy you say you have is kind of strange. Could you type the following command at a console while running as root and post the output.
locate run_manager
2. If you're using a desktop icon to start the BOINC manager, right click on the icon, click on properties, and see what command it executes. On my FC3 system, it's
/home/BOINC/run_manager
. That's the file you should edit. You should stop BOINC, edit the file, and restart BOINC by clicking on the icon.
3. Are you running the BOINC client as root or have you setup a separate user for it? If you're letting boincmgr start the BOINC client then, using whatever user you're logged in as when you run the boincmgr, type the following command and post the output. This will show the default limits for that user.
ulimit -a
4. I looked at the computer in your profile and it says you're running Linux 2.6.16.27.tex1.lve . Do you know if you have selinux enabled?
That should be enough to get started. I know there's some security limits on Linux 2.6.x about changing values with ulimit. I'll have to find where to change that again. There's a soft and a hard limit and then there's a file somewhere that let's you override that for certain users. I ran across some info on that the other day, but will have to find it again.
Regards,
-- David
EDIT: Left something out of one question and deleted quoted text since it was large. See the message this is in response to if you need to reference it.
David Bell,
Thank you very much for this excellent post and your Understanding.Just worked the Night shift at the cities newspaperm and am real "sore" to say the least.
Tommorrow morning, Thursday will find out all the info you are asking.I run Boinc in "user" not root.I can click the Boinc, then Boinc Manger in Home file system to get her running.But, I use the quick browrser which is just like Windows with the Start button.I will take some screen shots, of the file system,
The quick browser etc..I donnot know where I could post it unless at my Team
Test post.Team synergy, will give you link to look at screen shots, and we can use Docking, this thread to communicate.I hope the time you spend helping me get going again, doesnot affect your own crunching and stuff,Wow been awake 24 hours now, must get a few hours shut eye.If you donnot like the idea of screen shots
please let me know.run_manager is a text file, yes, the other will go look.It is a Back-up file, go figure, text also.Maybe I must do the ulimit in both?But before that will get those outputs for you.
Again, Thank you so much.
Sincerely
Doug Worrall Distro is "Big Daddy" came out 3 months ago, was running MiniMe
same thing, sort of From PCLINUXOS This will work
I just wanted to give everyone an update from the project side. I am currently working on adding code to change the stack size limit from the application. This is not an ideal solution because it may not work for everyone depending on your system settings.
We are also looking at the possibility of building a smaller version of the charmm application. This may take some time, because we need to make sure the smaller version works with our input files, and does indeed fix the stack problem.
Both of these are short term solutions. In the long term, we are working on moving to a newer version of the charmm application, but obviously this will take more work, and time, on our side.
Hello David,
Here is the output from run_manager in Root
bash-3.00$ su
Password:
[root@localhost doug]# locate run_manager
warning: locate: warning: database /var/lib/slocate/slocate.db' is more than 8 days old
/home/doug/BOINC/run_manager
[root@localhost doug]#
There are screen shots of my Files at the URL above
First, I'm assuming from the info that I saw that the user you're running BOINC under is "doug".
Here's something that might work.
go back to the original run_client and run_manager without the ulimit in them
Stop boinc and logout
login as root
CAREFULLY edit file /etc/security/limits.conf
At the bottom of the file, just before the line that says
# End of file
insert a line that says
doug soft stack 64000
Now, save the file and exit the editor. Note that there are some examples in that file that are commented out by having a '#' character in the first column. You don't want to put a hash mark in front of this line since that would make it a comment, to be ignored. The line should begin in column one, exactly as typed above.
logout as root and login as doug .
Start boinc the way you usually do. Any programs run by user doug should have a default stack limit of 64000 * 1024 (62.5 megabytes),which is probably enough.
Let me know if this works.
Could you open a command line while running as doug and type the following command and post the results so I can check them.
Thanks Dave,
Just on my way to work.Will do that Positive sounding and intelligent setting of my P.C.Do you mind if I paste this Post at my Forum?
Will inform you tommorrow of outcome, will do a "Top" in Terminal and watch it crunch, will see it also on Gkreml.
Thanks Dave,
Just on my way to work.Will do that Positive sounding and intelligent setting of my P.C.Do you mind if I paste this Post at my Forum?
Will inform you tommorrow of outcome, will do a "Top" in Terminal and watch it crunch, will see it also on Gkreml.
Sincerely
Doug*
Hello,
It's OK to post it somewhere else. Just put a caveat that they're trying it at their own risk.
Your linux kernel is a 2.6.x variant. I don't know if it will work with 2.4.x kernels or RHEL3 yet. IIRC, Redhat backports some things so even though RHEL3 is based on 2.4.x kernel, it may have features from 2.6.x kernels backported.
I'm beginning to wonder if some BOINC projects are going to hit other limits besides stack. IIRC, the default memory limit for a user on FC3 is 200 Megs and I know I've seen some BOINC apps come close to that. I haven't tried the Seasonal attribution project, but it recommends machines with at least a GB of RAM so I'm wondering if it goes over the default limit on some linux machines? I'm not running it since two of my machines are already running the ClimatePrediction 160 year model that takes about 4000 hours to complete, along with other BOINC projects such as Rosetta, which has the largest resource share, Einstein, Seti, SIMAP (on one machine instead of climate prediction) and some projects which don't have work, such as LHC and Predictor. Two machines are now running Docking. My machines are a celeron 2.3 Ghz/1.5 GB ram/WinXP, a celeron 2.4 Ghz/ 2GB ECC ram/ Linux, and a Socket A Sempron 2500+/ 1 GB ram/ Linux. They're not the fastest machines around, but they get the job done and are responsive. I'm not a gamer so I don't need super fast graphics. I would love to have a dual core machine for multi-threaded software testing, though.
Hello Dave,
Did not work. errored out.Did the command line for you anyways.
ash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -a
core file size (blocks, -c) 0
data seg size (kbytes, -d) unlimited
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 32
max memory size (kbytes, -m) unlimited
open files (-n) 1024
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) 64000
cpu time (seconds, -t) unlimited
max user processes (-u) 6073
virtual memory (kbytes, -v) unlimited
bash-3.00$
Hope "this" will help you understand {why} it did not work.The w/u errored out right at the end as per usual. Thanks for your help, seemed like this would work.
Hello Dave,
Did not work. errored out.Did the command line for you anyways.
ash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -a
core file size (blocks, -c) 0
data seg size (kbytes, -d) unlimited
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 32
max memory size (kbytes, -m) unlimited
open files (-n) 1024
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) 64000
cpu time (seconds, -t) unlimited
max user processes (-u) 6073
virtual memory (kbytes, -v) unlimited
bash-3.00$
Hope "this" will help you understand {why} it did not work.The w/u errored out right at the end as per usual. Thanks for your help, seemed like this would work.
Sincerely
Doug*
Actually, it did work as far as changing the stack size. If you look at stack size in the output from "ulimit -a", you'll see that it did change to 64000, which is the number we put in the file. I guess we'll just have to give user doug a much larger stack size.
Stop boinc and logout
login as root
CAREFULLY edit file /etc/security/limits.conf
At the bottom of the file, just before the line that says
# End of file
edit the line we put in that says
doug soft stack 64000
and change it to read
doug soft stack unlimited
doug hard stack unlimited
Note that I changed the number from 64000 to unlimited and added another line setting the hard limit to unlimited as well.
Now, save the file and exit the editor. Note that there are some examples in that file that are commented out by having a '#' character in the first column. You don't want to put a hash mark in front of these two lines since that would make each of them a comment, to be ignored. The lines should each begin in column one, exactly as typed above.
logout as root and login as doug .
Start boinc the way you usually do.
Abort ALL Docking@home workunits
and let it get new ones so that we're sure you have the latest charmm software.
Any programs run by user doug should now have a default stack limit of unlimited. If that doesn't fix it, then there's probably something else going on.
The next thing I would try would be to go to the Docking@home website, click on
your account
, edit your boinc
General preferences
and make sure that it's not exceeding the memory limit set there. Under the section titled "Disk and memory usage", tell it to use no more than
80% of total virtual memory
.
BTW, how much memory and swap space does your Linux machine have?
Hello Dave,
Took a bit figuring why when I entered command into that file, it left periods
so copied and pasted it in and no periods.Will do the new way as you have sugested
setting the soft and hard stacksize.Will delete what I put in yesterday and will put the new 2 lines in.Also will do the ulimit -a after I change it .
Here is my swap space and Ram
CPU type GenuineIntel
Intel(R) Celeron(R) CPU 2.40GHz
Number of CPUs 1
Operating System Linux
2.6.16.27.tex1.lve
Memory 749.29 MB
Cache 256 KB
Swap space 1592.34 MB
Total disk space 18.62 GB
Free Disk Space 17.74 GB
Measured floating point speed 533.52 million ops/sec
Measured integer speed 1644.33 million ops/sec
Hello Dave,
Took a bit figuring why when I entered command into that file, it left periods
so copied and pasted it in and no periods.Will do the new way as you have sugested.Also,
setting the soft and hard stacksize.Will delete what I put in yesterday and will put the new 2 lines in.Also will do the ulimit -a after I change it .
Here is my swap space and Ram
CPU type GenuineIntel
Intel(R) Celeron(R) CPU 2.40GHz
Number of CPUs 1
Operating System Linux
2.6.16.27.tex1.lve
Memory 749.29 MB
Cache 256 KB
Swap space 1592.34 MB
Total disk space 18.62 GB
Free Disk Space 17.74 GB
Measured floating point speed 533.52 million ops/sec
Measured integer speed 1644.33 million ops/sec
Thanks again Dave
Sincerely
Doug*
"It crashed"
# /etc/security/limits.conf
#
#Each line describes a limit for a user in the form:
#
#<domain> <type> <item> <value>
#
#Where:
#<domain> can be:
# - an user name
# - a group name, with @group syntax
# - the wildcard *, for default entry
# - the wildcard %, can be also used with %group syntax,
# for maxlogin limit
#
#<type> can have the two values:
# - "soft" for enforcing the soft limits
# - "hard" for enforcing hard limits
#
#<item> can be one of the following:
# - core - limits the core file size (KB)
# - data - max data size (KB)
# - fsize - maximum filesize (KB)
# - memlock - max locked-in-memory address space (KB)
# - nofile - max number of open files
# - rss - max resident set size (KB)
# - stack - max stack size (KB)
# - cpu - max CPU time (MIN)
# - nproc - max number of processes
# - as - address space limit
# - maxlogins - max number of logins for this user
# - priority - the priority to run user process with
# - locks - max number of file locks the user can hold
#
#<domain> <type> <item> <value>
#
#* soft core 0
#* hard rss 10000
#@student hard nproc 20
#@faculty soft nproc 20
#@faculty hard nproc 50
#ftp hard nproc 0
#@student - maxlogins 4
doug soft stack unlimited
doug hard stack unlimited
# End of file
Hello, it crashed when I saved it, the crash output said to check my
Diskspace and Ram, I think, the crash always makes me Jump, Thos elightbulbs crashing are annoying, LOL.It asked me if I wanted to overwrite the file with
the above.I said not to overwrite due to the crash after I tried saving it?Also
the disk useage {by default} were set at 50%, changed to 80% now.
WOOPS,
Forgot to do it in Root, did it, will put the output from command now, D,oh
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -a
core file size (blocks, -c) 0
data seg size (kbytes, -d) unlimited
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 32
max memory size (kbytes, -m) unlimited
open files (-n) 1024
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) 8192
cpu time (seconds, -t) unlimited
max user processes (-u) 6073
virtual memory (kbytes, -v) unlimited
bash-3.00$
Hello Dave,
Just tried another Docking w/u after all the changes were made "correctly".Thanks
for all your help. I guess this is a problem with "my distro" and am at witts
end.Again thank you for all your help.I will try anything and will not give-up, should I go back to cofig file and delete changes?
Sincerely
Doug Worrall
ID:
707 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Doug: some questions
what distro do you have?
what shell are you using?
are you running boinc in graphical or text mode?
Doug: some questions
what distro do you have?
what shell are you using?
are you running boinc in graphical or text mode?
How long do your WUs take to exit with code 1?
Hello,
Am running "Big Daddy" Linuxos 93a 2 months ago released
am using Boinc 5.4.9 and donnot know if I am running in graphical or text mode?
and just this morning I wrote down the exact times , the w/u never make it past
3.475% but finishes with error code approximately 5 minutes 45 seconds
same as charm 2.01 which ran just fine untill they changed over too this one?
I hope this is of use for you .Sincerely
Doug Worrall
ID:
710 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
1. How about your shell? For this you can check the contents of /etc/passwd en at the end of the line of the account that you use to run boinc should say. Examples are bash (being one of the most common, I assume), zsh, ksh, csh, tcsh ,etc
2. Do you manipulate BOINC(attach to projects, suspend, resume, etc) with buttons or just send commands? (I am almost sure you are doing graphical)
I really hope we can find a temp fix for you, just work with me :)
1. How about your shell? For this you can check the contents of /etc/passwd en at the end of the line of the account that you use to run boinc should say. Examples are bash (being one of the most common, I assume), zsh, ksh, csh, tcsh ,etc
2. Do you manipulate BOINC(attach to projects, suspend, resume, etc) with buttons or just send commands? (I am almost sure you are doing graphical)
I really hope we can find a temp fix for you, just work with me :)
Thank You,
A temp fix will end this daily grind I have been struggling with for about 3 weeks now.It said I had no permission to read my passord file, at first, so I opened
the one that was a plain text file.I copied it:
bin:x:1:1:bin:/bin:/bin/sh
daemon:x:2:2:daemon:/sbin:/bin/sh
adm:x:3:4:adm:/var/adm:/bin/sh
lp:x:4:7:lp:/var/spool/lpd:/bin/sh
sync:x:5:0:sync:/sbin:/bin/sync
shutdown:x:6:0:shutdown:/sbin:/sbin/shutdown
halt:x:7:0:halt:/sbin:/sbin/halt
mail:x:8:12:mail:/var/spool/mail:/bin/sh
news:x:9:13:news:/var/spool/news:/bin/sh
uucp:x:10:14:uucp:/var/spool/uucp:/bin/sh
operator:x:11:0:operator:/var:/bin/sh
games:x:12:100:games:/usr/games:/bin/sh
nobody:x:65534:65534:Nobody:/:/bin/sh
rpm:x:13:101:system user for rpm:/var/lib/rpm:/bin/false
vcsa:x:69:69:virtual console memory owner:/dev:/sbin/nologin
xfs:x:70:70:system user for XFree86:/etc/X11/fs:/bin/false
postfix:x:71:71:system user for postfix:/var/spool/postfix:/bin/false
messagebus:x:72:72:system user for dbus:/:/sbin/nologin
haldaemon:x:73:73:system user for hal:/:/sbin/nologin
saned:x:74:74:system user for saned:/home/saned:/bin/false
sshd:x:75:75:system user for openssh:/var/empty:/bin/true
doug:x:501:501:doug:/home/doug:/bin/bash
regarding the way I run Boinc is with buttons, never use command lines till 3 weeks ago.Hope this is of use for you,I may have to go to Root to get another reading from this file.
Hello Dave,
Just tried another Docking w/u after all the changes were made "correctly".Thanks
for all your help. I guess this is a problem with "my distro" and am at witts
end.Again thank you for all your help.I will try anything and will not give-up, should I go back to cofig file and delete changes?
Sincerely
Doug Worrall
You probably should go back and remove the entries for doug from the limits.conf file to be safe. It's strange that the 64000 setting seemed to have worked for user doug (as shown by the ulimit -a output), but the unlimited was ignored. It sure looks like there is a lot of difference between Linux distros.
Unfortunately, I don't have a machine I could load your distro on and even if I did, I'm in a rural area which only has Dial-up internet out this far. Downloading CD images is beyond 56k Dial-up capabilities.
Wish I could be more help,
-- David
ID:
718 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
OK seems like you run bash
In your home directory you should have a file called .bashrc (note the leading dot!) If you add the <ulimit -s unlimited> at the end of the file and save
then on a shell prompt type:
bash
ulimit -s
what is the output?
I will try to install your distro on a virtual machine to better help you, but if this "sticks" it might get you up and running for now.
OK seems like you run bash
In your home directory you should have a file called .bashrc (note the leading dot!) If you add the <ulimit -s unlimited> at the end of the file and save
then on a shell prompt type:
bash
ulimit -s
what is the output?
I will try to install your distro on a virtual machine to better help you, but if this "sticks" it might get you up and running for now.
Hello,
Had to find .bashrc file in hidden files, typed and saved ulimit -s unlimited,
typed bash ulimit -s on a shell, got message "No such file in directory" Go Figure?Ran a w/u and got the usual
Sun Oct 1 09:28:09 2006|Docking@Home|Unrecoverable error for result 1tng_mod0001_4256_477205_3 (process exited with code 1 (0x1))
Sincerely
Had to find .bashrc file in hidden files, typed and saved ulimit -s unlimited,
typed bash ulimit -s on a shell, got message "No such file in directory" Go Figure?
It's separate commands. First type
bash
, press enter, then type
ulimit -s
.
Had to find .bashrc file in hidden files, typed and saved ulimit -s unlimited,
typed bash ulimit -s on a shell, got message "No such file in directory" Go Figure?
It's separate commands. First type
bash
, press enter, then type
ulimit -s
.
Thanks,
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ bash
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -s
unlimited
bash-3.00$
Now from that same commandline where you see 'unlimited' change directory into your BOINC installation (probably 'cd BOINC') and run the boincmgr (enter './boincmgr'). Make sure you stop any running boinc programs first before you start this one.
Andre
Had to find .bashrc file in hidden files, typed and saved ulimit -s unlimited,
typed bash ulimit -s on a shell, got message "No such file in directory" Go Figure?
It's separate commands. First type
bash
, press enter, then type
ulimit -s
.
Thanks,
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ bash
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -s
unlimited
bash-3.00$
Got "unlimited" yet still no go?
Sincerely
Doug*
____________
D@H the greatest project in the world... a while from now!
ID:
738 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
To make sure that no other threads of boinc are not running you can check with
"ps -aux"
and see if there is a process that says charmm or boinc
to kill it right away get the PID of that process and do
kill -9 <PID>
Do that for each process that you see that says charmm or boinc and then restart the project again.
NOTE:
By killing boinc you will suspend any other projects that you are running but they will be resumed when you rerun boinc.
hello,Me Again
I will never give in, have been told that this newKernel of mine, has security
measures of some sort?I am a newbie, I am wrong probably?Here is the Output from
./boincmngr
ash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ bash
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -s
unlimited
bash-3.00$ cd./boincmgr
bash: cd./boincmgr: No such file or directory
bash-3.00$
Albeit, seem to get further each time
MEMO, regarding your "post", I apologise, but, I donnot understand what that
procedure is for?
hello,Me Again
I will never give in, have been told that this newKernel of mine, has security
measures of some sort?I am a newbie, I am wrong probably?Here is the Output from
./boincmngr
ash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ bash
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ulimit -s
unlimited
bash-3.00$ cd./boincmgr
bash: cd./boincmgr: No such file or directory
bash-3.00$
Albeit, seem to get further each time
MEMO, regarding your "post", I apologise, but, I donnot understand what that
procedure is for?
Sincerely
Doug Worrall
4 weeks trying this Fix {smile}
What do you have on your ./boincmgr? looks like you typed literally [doug@localhost] on it, so bash is interpreting it as a command...
ID:
791 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Also if its already running you can use:
nohup ./run_client &
This will run it in the background, if you want to see the output it will be stored in a file called nohup.out
just remember to run this inside the BOINC directory
I hope this helps
I downloaded a copy of your distro I will be installing this during the weekend to help you better.
security issues with the linux kernel is something that happens all the time on all the operative systems. With this in mind I can tell you that I am 99.9% sure that it doesn't affect anything in BOINC.
Thanks Memo,
This whole "problem" has left me with no way, unless you can find out the way with this Distro,Much appreciated,
Doug
Hello Memo,
Through "Symnaptic"{Yum} just updated whole system.#1=new KernelOperating System Linux
2.6.16.29.tex1.lve
#2=KDE updated too 3.5.5.
#3= I have a miriad of Kernels I can choose from, will try this new one, then was thinking about trying an i686 High Memory Kernel, wheras the older Kernel 2.6.16.27 is from Live c/d install, will "try" to get some info from the pclinuxos
Boards for some help, have tried before, and got "What the heck is Boinc?"
never had RTFM, yet, cause with this Distro, "The Manual"is Archaic and not for newbies. I am hoping the new kernel might work.Am patiently waiting to here back from "Yall"
Sincerely
Doug Worrall
I'm, very sorry I can't read all the thread (I read a big part of it), so I'll learn it from you. It seems that you run a Mandriva-derived distro. I'm running a debian based distro and the ulimit command is working fine, but since we have different init scripts for the client, I have to learn how your script works (but I'd suggest to use a debian based distro all the way).
So, using a Mandriva-like (or Fedora Core like?) distro your client is running with doug user's permissions. I suppose doug is your main user.
If doug is your main user, absolutely DON'T give him unlimited stack in the limits.conf file, since this could be the cause of greater problems with potentially buggy apps (not necessarily docking app). If your user has unlimited stack all the way, every process started by a shell instance from doug will be potentially dangerous for your system.
So, let's understand what it happens when your system boots, by answering a question at a time (I'm not an expert in Mandriva or FC based distros, I'm in love with debian).
Do you have the client started at boot time or you have to start it manually?
This is the first fundamental thing to know. When you correctly modified the limits.conf file, the app crashed as well, is it right? Which number of minutes did it take to crash? The same number as with the original stack size?
A question: the word "FIX" in the title is to say that the problem has been fixed by an upgrade from your distro's mantainers?
Uhmmm, maybe you would have written FIXED, so I prefer to ask.
If this is the case, which size have you choose for the stack?
Anyway, it's absolutely not safe to give your main user (doug) the permission to start unlimited stack processes. So it's far better to modify the run_client script than adding a row for doug in limits.conf.
A question: the word "FIX" in the title is to say that the problem has been fixed by an upgrade from your distro's mantainers?
Uhmmm, maybe you would have written FIXED, so I prefer to ask.
If this is the case, which size have you choose for the stack?
Anyway, it's absolutely not safe to give your main user (doug) the permission to start unlimited stack processes. So it's far better to modify the run_client script than adding a row for doug in limits.conf.
Hello Danilele,
Thanks for your reply to this thread. After 5 hours of "ulimit -s unlimited
has Fubarred {I have fubarred} My Boinc Manager, and now, cannot get a trickle
to leave for Climate.I start Boinc after Boot, by using my "Quick Browser" into your Home Folder where Boinc is presently.Not in Root.I hit the Boinc first,
then Boinc Manager, have been doing this for close to 2 years without any Fubarred
w/u.Actually come close to Dual Cores in times to finish somw w/u.Definetly a great Distro {for me} This is the Url for My system
http://www.pclinuxos.com/forum/index.php?topic=9076.0
Be Well Danilelle
Sincerely
Doug*
Hi Doug :)
I'll try to help you, but the problem seems to lie in your distro structure, and I'm sorry I'm not an expert in Mandriva or Fedora's structure...
So, let me see what there's in that file, run_client.
Type
cd /home/doug/BOINC/
cat run_client
and post the output.
A question: at this moment, are you using `ulimit -s ...` somewhere in your system or you returned to the original options?
I don't want to deceive you, but there's a chance that your problem is solvable in few minutes, if it stays the way I suppose. If not, then it's a distro's problem. But before going on, I'll wait for your answers.
I don't want to deceive you, but there's a chance that your problem is solvable in few minutes, if it stays the way I suppose. If not, then it's a distro's problem. But before going on, I'll wait for your answers.
Hello Danniele,
Its telling me that there is "no such file in directory" I assume the / are spaces
for the command I typed twice.Also I have set stack size back to 8192 by ridding
the saved Ulimit command.
Sincerely
Doug*
[quote]Its telling me that there is "no such file in directory" I assume the / are spaces
for the command I typed twice.Also I have set stack size back to 8192 by ridding
the saved Ulimit command.
I don't understand, why should be the / spaces? :)
The / are /, everything is exactly what it seems, I'll be very precise to reach the solution to your problem in little time, if it exists (and I'm pretty sure it does). I also want you to understand what you are doing, so I'll tell exactly what to do, to avoid misunderstanding.
So, I saw in that link (you posted) that you have boinc's files in /home/doug/BOINC. Aren't they?
/ is the charecter to divide directories in a path, `cd` is the command to "go" to a directory (maybe "Current Directory"? I don't remember).
cd /home/doug/BOINC
means "go to the directory BOINC, which is in directory "doug", which is in directory "home", which is in directory / (the root directory, it's a special one).
Once you are there,
cat run_client
This command will write what's in the file on the screen.
Then you copy the output and post it here :)
I have a basic quastion.
When you start the client, what link you click on, exactly?
The one named 'boinc'?
This would be an error, probably.
ID:
1084 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
[quote]Its telling me that there is "no such file in directory" I assume the / are spaces
for the command I typed twice.Also I have set stack size back to 8192 by ridding
the saved Ulimit command.
I don't understand, why should be the / spaces? :)
The / are /, everything is exactly what it seems, I'll be very precise to reach the solution to your problem in little time, if it exists (and I'm pretty sure it does). I also want you to understand what you are doing, so I'll tell exactly what to do, to avoid misunderstanding.
So, I saw in that link (you posted) that you have boinc's files in /home/doug/BOINC. Aren't they?
/ is the charecter to divide directories in a path, `cd` is the command to "go" to a directory (maybe "Current Directory"? I don't remember).
cd /home/doug/BOINC
means "go to the directory BOINC, which is in directory "doug", which is in directory "home", which is in directory / (the root directory, it's a special one).
Once you are there,
cat run_client
This command will write what's in the file on the screen.
Then you copy the output and post it here :)
I have a basic quastion.
When you start the client, what link you click on, exactly?
The one named 'boinc'?
This would be an error, probably.
I think that clicking boinc means go to boinc directory
[quote]Its telling me that there is "no such file in directory" I assume the / are spaces
for the command I typed twice.Also I have set stack size back to 8192 by ridding
the saved Ulimit command.
I don't understand, why should be the / spaces? :)
The / are /, everything is exactly what it seems, I'll be very precise to reach the solution to your problem in little time, if it exists (and I'm pretty sure it does). I also want you to understand what you are doing, so I'll tell exactly what to do, to avoid misunderstanding.
So, I saw in that link (you posted) that you have boinc's files in /home/doug/BOINC. Aren't they?
/ is the charecter to divide directories in a path, `cd` is the command to "go" to a directory (maybe "Current Directory"? I don't remember).
cd /home/doug/BOINC
means "go to the directory BOINC, which is in directory "doug", which is in directory "home", which is in directory / (the root directory, it's a special one).
Once you are there,
cat run_client
This command will write what's in the file on the screen.
Then you copy the output and post it here :)
I have a basic quastion.
When you start the client, what link you click on, exactly?
The one named 'boinc'?
This would be an error, probably.
I think that clicking boinc means go to boinc directory
cd means Change Directory,
/// did it with and witout /// these."NO Such File In Directory"
Doug
Thanks
Did you see the screen shots of Boinc Now, sems evryone else,staff, scientists, including "self" has
given-up.Every other w/u from every other experiments work, assides from UTEP,
seeing as Simap is out of the game , sure would like a Docking w/u not to error out.My
Kernel" is new and safe and unable to change things with security features.Should not have to too fascilatate w/u that do not FUbar.If i hit Boinc Manager first, the que is emty and it benches and asks to hook to experiments, that is why you hit "Boinc" first then Manager!
Sincerely
Doug
Did you see the screen shots of Boinc Now, sems evryone else,staff, scientists, including "self" has
given-up.Every other w/u from every other experiments work, assides from UTEP,
seeing as Simap is out of the game , sure would like a Docking w/u not to error out.My
Kernel" is new and safe and unable to change things with security features.Should not have to too fascilatate w/u that do not FUbar.If i hit Boinc Manager first, the que is emty and it benches and asks to hook to experiments, that is why you hit "Boinc" first then Manager!
Let's suppose you have that file. Why are you clicking on the boinc link and not on the runc_client script? I've read something about your boinc package and I found that you have to use run_client. Most of the times it's not an optional. Bacause of this I want to look at run_client, study it and maybe modify it, then make you use it.
Have you ever used that file?
If I read well, in the forst page of this thread you said that you have inserted `ulimit -s unlimited` in the run_client file, but everything stayed he same. The problem is: if you click on boinc and not on run_client, run_client script will be ignored. I wonder why nobody here has told you.
That is not a simple text file, it's a script.
Said this, I'm sorry if I made mistakes in understanding your words, but it's often hard to me since I'm not english and sometimes your speaking is too difficult or complicated to me.
Did you see the screen shots of Boinc Now, sems evryone else,staff, scientists, including "self" has
given-up.Every other w/u from every other experiments work, assides from UTEP,
seeing as Simap is out of the game , sure would like a Docking w/u not to error out.My
Kernel" is new and safe and unable to change things with security features
Now I'm even more serious than before.
In these two days I read about your distribution, and I think it's very nice, coloured, simple. Good to start, indeed. Your desktop is very nice as well.
But that's not linux, if you can't change processes' permissions, if you have to give up for such a simple matter.
My (serious) advice is: use Kubuntu if you have a good pc. Use a debian based and you will never have such these problems Giving up doesn't exist if the distro has a good structure. Kubuntu is pretty easy if you want it to be easy, and it's debian in its core.
Anyway, if you don't want to use a distro which permits to really manage the system, then we'll get to point with Big Daddy, if it has not any bugs inside the security substructure.
got this from run manager
cd "/home/doug/BOINC" && exec ./boincmgr $@
My Linux expert was over a month ago twice, trying to figure this out.I just retyped in ulimit -s, and got "Unlimited", so it took when I first did it, yet,
maybe I have been messing-up trying?
Thank-you for your advice
Sincerely
Doug
got this from run manager
cd "/home/doug/BOINC" && exec ./boincmgr $@
My Linux expert was over a month ago twice, trying to figure this out.I just retyped in ulimit -s, and got "Unlimited", so it took when I first did it, yet,
maybe I have been messing-up trying?
Thank-you for your advice
Sincerely
Doug
I'm sorry again I can't understand your english perfectly :(
But, for sure, the row
cd "/home/doug/BOINC" && exec ./boincmgr $@
is perfectly clear.
Though, the problem is not the manager, but the client.
I need to know what's in the run_client script, to understande if it' necessary (I think it is, since in the README file of the package there's written that you HAVE to use it, and also in debian I use a script to make the client work. Without that script, the client doesn't work fine.)
Hello,
I can "easily" redo this disto from scatch 15 minutes, then the Boinc install would be "fresh" for either Slax, Damn small Linux, Mini Me, PCLINUXOS 9.2 which is big daddy now, or, Mandriva 2006, Ubuntu 5.10, MiniMe and Big Daddy are my preferences.
The instuctions for installing Boinc are REAL "Vague", at the Posts for this Distro, not 1 person understands, or, heard of it {BOINC} .Have even touched base with tex,
and he "heard" of Boinc, but new nada.I understand our Communication difficulties due too Language, and the innability to see my system "sight unseen",As you can see above, the forum Moderators were going to put this Distro in a "virtual Machine", or something.Am still waiting for advice regarding the Virtual machine.
I may do a complete reinstall, I hate to lose another Climate CHUNK, but, getting
"that error in screen shots" for Boinc Manager now, no trickles from Climate in a week, soo, a install takes about 15 minutes then to tweek, about another hour?
Regarding the scipt for run_client for Debian? could not be same as this Disto?
Sincerely
Doug*
Regarding the scipt for run_client for Debian? could not be same as this Disto?
Debian package doesn't have the scripts run_client and run_manager, but much better scripts to run safely the client with a separate user (boinc user).
So I cannot use my script for resolving your problem, I need to look at your script.
So, if you can do exactly what I wrote before, looking at the screenshot I can see the file, why cannot you as well?
My advice is the same: use Kubuntu, and NOT use the package from boinc, but the one from Kubuntu (debian) repository.
That command applies ONLY to a shell instance, if you make the core start by clicking on a link in kde you WON'T use the unlimited stack. It's a bit difficult to explain since I'm not with you near your host, but the only way to make the `ulimit` work is to type it in a shell, then starting the CLIENT (NOT the manager, the manager is NOTHING to the client) in that shell somehow.
You said you typed `ulimit` and then started the manager. Ok, but the manager is not your problem at this moment. The problemis the client.
So, if you type `ulimit` in a shell then you click on the kde link to the client, the `ulimit` command is NO USEFUL, the client won't see it.
You have (it's what I'm trying to make you do) to modify "run_client" script, then USE THE SCRIPT. If this works, then we'll create a link in kde to make it use the script in the right way. It's not difficult, this is the only way to make the `ulimit` command work.
Regarding the scipt for run_client for Debian? could not be same as this Disto?
Debian package doesn't have the scripts run_client and run_manager, but much better scripts to run safely the client with a separate user (boinc user).
So I cannot use my script for resolving your problem, I need to look at your script.
So, if you can do exactly what I wrote before, looking at the screenshot I can see the file, why cannot you as well?
My advice is the same: use Kubuntu, and NOT use the package from boinc, but the one from Kubuntu (debian) repository.
Thank you for your advice,
I can see the file you speek of, yet it has "no affect" when I use it?
My Repos are Symnaptic, not yum, nor do I have the repos for Debian.
Be Well
Doug
Besides this, have you EVER tried something like this?
1) modifying run_client by adding ulimit -s unlimited as the first row
2) typing in a terminal (being in the /home/doug/BOINC directory)
./run_client
or at least clicking on the link to run_client to make it start
Besides this, have you EVER tried something like this?
1) modifying run_client by adding ulimit -s unlimited as the first row
2) typing in a terminal (being in the /home/doug/BOINC directory)
./run_client
or at least clicking on the link to run_client to make it start
Have you ever tried?
Yes,
I jave tryed all that.The first line reads "Ulimited" ever sice I did the ulimit -s {Fix}.Maybe I am doing something wrong with start-up, but the only experiment
that does not work, is this one.Have tried all there is to try the "Fix" it worked,
reset and unreset the stacksize etc.....................................................................etc.......................
Really am not "understanding" what the please,etc.. is about, but this is over for me.I can run anyother experiment with great credits for Linux, and am not concerned regarding Docking.Have been crunching flawlessly assides "Docking".
Thank you for your concern, but, I donnot have the time for ONE Exeriment
that does not work."So Be It"
Sincerely
"over and OUT"
Sluger
Really am not "understanding" what the please,etc.. is about, but this is over for me.I can run anyother experiment with great credits for Linux, and am not concerned regarding Docking.Have been crunching flawlessly assides "Docking".
Thank you for your concern, but, I donnot have the time for ONE Exeriment
that does not work."So Be It"
Sincerely
"over and OUT"
Sluger
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thoughy for a moment you wanted to help and to improve your know-how, but probably this is too diffcult for your, at this moment.
Thanks for your help Daniele. It's okay if the ulimit fix doesn't work for some people. We are working on another more permanent solution and when that is released, Charmm will hopefully work for all users.
Thanks for your help again! Much appreciated.
Andre
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thoughy for a moment you wanted to help and to improve your know-how, but probably this is too diffcult for your, at this moment.
Have a good time crunching.
PS: docking@home IS working, fine
____________
D@H the greatest project in the world... a while from now!
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thought for a moment you wanted to help and to improve your know-how, but probably this is too difficult for your, at this moment.
Have a good time crunching.
PS: docking@home IS working, fine
We couldn't help Doug, but maybe you can help me get it working on one of my machines. I have it working on Fedora Core 3, but can't get it to run on a RedHat RHEL3 machine that is command line only. The kernel is (uname -a)
Linux xxxxxxxxxxxx 2.4.21-47.0.1.EL #1 Fri Oct 13 18:04:55 EDT 2006 i686 i686 i386 GNU/Linux
Here's the last part of the section of my /etc/init.d/boinc that starts the client.
cd $BOINCDIR
if [ -f lockfile ] ; then
echo -n "Another instance of BOINC is running (lockfile exists)."
echo_failure
echo
exit 4
fi
if [ ! -f client_state.xml ] ; then
echo -n "The BOINC client requires initialization."
echo_warning
echo
fi
echo -n "Starting BOINC client as a daemon: "
ulimit -s 199000
ulimit -d 512000
ulimit -v 512000
su $BOINCUSER -c "ulimit -a" >>$LOGFILE 2>>$ERRORLOG
su $BOINCUSER -c "nice --adjustment=19 $BOINCEXE $BOINCOPTS" >>$LOGFILE 2>>$ERRORLOG &
sleep 1
PID=`pidof -s -x -o $$ -o $PPID -o %PPID $BOINCEXE`
if [ $PID ]; then
touch $LOCKDIR/boinc
echo_success
else
echo_failure
fi
echo
;;
Increasing the stack size did seem to stick and after this, the run time for a Docking@Home WU still failed after about 400 seconds with Exit status : 1 (0x1). The new longer Work Units run about 750 seconds before getting the same error.
Here's the output from the "ulimit -a" in the boinc logfile from the first su command, followed by the startup of boinc 5.2.13. I'm still on 5.2.13 because I haven't figured out how to change the init script to run BOINC 5.4.9 as a service, but that's a separate issue.
core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 512000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 4
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) 199000
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 512000
2006-10-25 19:50:12 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-10-25 19:50:12 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-10-25 19:50:12 [---] Data directory: /home/BOINC
2006-10-25 19:50:12 [---] Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
2006-10-25 19:50:12 [---] Memory: 1.95 GB physical, 1.95 GB virtual
2006-10-25 19:50:12 [---] Disk: 16.02 GB total, 11.69 GB free
2006-10-25 19:50:12 [Einstein@Home] Computer ID: 544961; location: work; project prefs: default
2006-10-25 19:50:12 [SETI@home] Computer ID: 2185126; location: work; project prefs: default
2006-10-25 19:50:12 [lhcathome] Computer ID: 113223; location: work; project prefs: default
2006-10-25 19:50:12 [rosetta@home] Computer ID: 211470; location: work; project prefs: work
2006-10-25 19:50:12 [Docking@Home] Computer ID: 223; location: work; project prefs: default
2006-10-25 19:50:12 [---] General prefs: from rosetta@home (last modified 2006-10-22 08:20:21)
2006-10-25 19:50:12 [---] General prefs: no separate prefs for work; using your defaults
2006-10-25 19:50:12 [---] Remote control not allowed; using loopback address
2006-10-25 19:50:14 [---] Running CPU benchmarks
2006-10-25 19:51:13 [---] Benchmark results:
2006-10-25 19:51:13 [---] Number of CPUs: 1
2006-10-25 19:51:13 [---] 645 double precision MIPS (Whetstone) per CPU
2006-10-25 19:51:13 [---] 1234 integer MIPS (Dhrystone) per CPU
2006-10-25 19:51:13 [---] Finished CPU benchmarks
2006-10-25 19:51:14 [rosetta@home] Resuming computation for result 1ogw__LARS_ABRELAX_PAIR5_BARCODE__1294_4165_0 using roset$
2006-10-25 19:51:14 [Einstein@Home] Deferring computation for result l1_0343.5_S5R1__5823_S5R1a_0
2006-10-25 19:51:14 [---] Resuming computation and network activity
2006-10-25 19:51:14 [---] request_reschedule_cpus: Resuming activities
Any guesses what else needs to be changed?
Once it's running, I control it by logging in as a regular user, doing "su -" to switch to root (I have ssh configured to not allow direct root login), then as root I do "su boinc" and "cd ~" which puts me in the correct directory. After that I use ./boinc_cmd to control the boinc client running as a service. Other projects work fine, but I've never gotten Docking@home to work, even though the "ulimit -s 199000" in the init.d script does seem to work.
Have you tried with ulimit -s unlimited? I tested several stack limits, but only unlimited seems to go through. Let me know if that still gives you problems. If it does, we might be looking at another issue.
Thanks
Andre
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thought for a moment you wanted to help and to improve your know-how, but probably this is too difficult for your, at this moment.
Have a good time crunching.
PS: docking@home IS working, fine
We couldn't help Doug, but maybe you can help me get it working on one of my machines. I have it working on Fedora Core 3, but can't get it to run on a RedHat RHEL3 machine that is command line only. The kernel is (uname -a)
Linux xxxxxxxxxxxx 2.4.21-47.0.1.EL #1 Fri Oct 13 18:04:55 EDT 2006 i686 i686 i386 GNU/Linux
Here's the last part of the section of my /etc/init.d/boinc that starts the client.
cd $BOINCDIR
if [ -f lockfile ] ; then
echo -n "Another instance of BOINC is running (lockfile exists)."
echo_failure
echo
exit 4
fi
if [ ! -f client_state.xml ] ; then
echo -n "The BOINC client requires initialization."
echo_warning
echo
fi
echo -n "Starting BOINC client as a daemon: "
ulimit -s 199000
ulimit -d 512000
ulimit -v 512000
su $BOINCUSER -c "ulimit -a" >>$LOGFILE 2>>$ERRORLOG
su $BOINCUSER -c "nice --adjustment=19 $BOINCEXE $BOINCOPTS" >>$LOGFILE 2>>$ERRORLOG &
sleep 1
PID=`pidof -s -x -o $$ -o $PPID -o %PPID $BOINCEXE`
if [ $PID ]; then
touch $LOCKDIR/boinc
echo_success
else
echo_failure
fi
echo
;;
Increasing the stack size did seem to stick and after this, the run time for a Docking@Home WU still failed after about 400 seconds with Exit status : 1 (0x1). The new longer Work Units run about 750 seconds before getting the same error.
Here's the output from the "ulimit -a" in the boinc logfile from the first su command, followed by the startup of boinc 5.2.13. I'm still on 5.2.13 because I haven't figured out how to change the init script to run BOINC 5.4.9 as a service, but that's a separate issue.
core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 512000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 4
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) 199000
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 512000
2006-10-25 19:50:12 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-10-25 19:50:12 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-10-25 19:50:12 [---] Data directory: /home/BOINC
2006-10-25 19:50:12 [---] Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
2006-10-25 19:50:12 [---] Memory: 1.95 GB physical, 1.95 GB virtual
2006-10-25 19:50:12 [---] Disk: 16.02 GB total, 11.69 GB free
2006-10-25 19:50:12 [Einstein@Home] Computer ID: 544961; location: work; project prefs: default
2006-10-25 19:50:12 [SETI@home] Computer ID: 2185126; location: work; project prefs: default
2006-10-25 19:50:12 [lhcathome] Computer ID: 113223; location: work; project prefs: default
2006-10-25 19:50:12 [rosetta@home] Computer ID: 211470; location: work; project prefs: work
2006-10-25 19:50:12 [Docking@Home] Computer ID: 223; location: work; project prefs: default
2006-10-25 19:50:12 [---] General prefs: from rosetta@home (last modified 2006-10-22 08:20:21)
2006-10-25 19:50:12 [---] General prefs: no separate prefs for work; using your defaults
2006-10-25 19:50:12 [---] Remote control not allowed; using loopback address
2006-10-25 19:50:14 [---] Running CPU benchmarks
2006-10-25 19:51:13 [---] Benchmark results:
2006-10-25 19:51:13 [---] Number of CPUs: 1
2006-10-25 19:51:13 [---] 645 double precision MIPS (Whetstone) per CPU
2006-10-25 19:51:13 [---] 1234 integer MIPS (Dhrystone) per CPU
2006-10-25 19:51:13 [---] Finished CPU benchmarks
2006-10-25 19:51:14 [rosetta@home] Resuming computation for result 1ogw__LARS_ABRELAX_PAIR5_BARCODE__1294_4165_0 using roset$
2006-10-25 19:51:14 [Einstein@Home] Deferring computation for result l1_0343.5_S5R1__5823_S5R1a_0
2006-10-25 19:51:14 [---] Resuming computation and network activity
2006-10-25 19:51:14 [---] request_reschedule_cpus: Resuming activities
Any guesses what else needs to be changed?
Once it's running, I control it by logging in as a regular user, doing "su -" to switch to root (I have ssh configured to not allow direct root login), then as root I do "su boinc" and "cd ~" which puts me in the correct directory. After that I use ./boinc_cmd to control the boinc client running as a service. Other projects work fine, but I've never gotten Docking@home to work, even though the "ulimit -s 199000" in the init.d script does seem to work.
Any Suggestions?
-- David
____________
D@H the greatest project in the world... a while from now!
Really am not "understanding" what the please,etc.. is about, but this is over for me.I can run anyother experiment with great credits for Linux, and am not concerned regarding Docking.Have been crunching flawlessly assides "Docking".
Thank you for your concern, but, I donnot have the time for ONE Exeriment
that does not work."So Be It"
Sincerely
"over and OUT"
Sluger
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thoughy for a moment you wanted to help and to improve your know-how, but probably this is too diffcult for your, at this moment.
Have a good time crunching.
PS: docking@home IS working, fine
Tut, Tut,
Sticks and stones, I am testing Alpha experiments, am still hooked to Docking, cause as I said Dannielle, I am reaapplying this system.Maybe I donot know it that well, came out only 3 months prior.I thanked you for your help, yet you feel the need to call names.Maybe you donnot know "Yourself" well enough
to know that I have been at this for 35 days, and some nights.Also as I mentioned, the language problem some stuff I donnot understand what you are saying."I will get this working by hook or by crook".Just not as I have cemotherapy today and fel like hell warmed over.
Dannilele, again, thank you for your "help"Also, I refuse to give-up, each week is cemotherapy, today was a day and feel quite yukky, this is why I like to run folding experiments, am sure if anyone tried to make their Distro take a charm w/u, last one yesterday, errored out after I tried new things for 40 days would "give-up"I cannot, but for today I have.Again, language barrier neccesatated you to verbally attack me.tut, tut, where are your manners?
.{ANDRE}, I asked you to "Please" remove that taking
back 200% share, obviosly Dannile has read that.Good luck with the machine to work for you and hope Dannielle can help you.Thank- you again Danniele,Maybe I should learn some Italian, then communication "may" get better, again Danniele,
Be Well
Sincerely
Doug
Also Dave Bell,
Thanks for your help also.Good distro,Hope you can get her running for Docking.
Best wishes also Andre
Have you tried with ulimit -s unlimited? I tested several stack limits, but only unlimited seems to go through. Let me know if that still gives you problems. If it does, we might be looking at another issue.
Thanks
Andre
OK, I changed /etc/init.d/boinc to set the stack to unlimited and I doubled the memory values to about 1 GB. Here's the boinc.log from where I started the service up again (as dumped from nano so long lines end in $).
core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 1024000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 4
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 1024000
2006-10-26 20:17:09 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-10-26 20:17:09 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-10-26 20:17:09 [---] Data directory: /home/BOINC
2006-10-26 20:17:09 [---] Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
2006-10-26 20:17:09 [---] Memory: 1.95 GB physical, 1.95 GB virtual
2006-10-26 20:17:09 [---] Disk: 16.02 GB total, 11.74 GB free
2006-10-26 20:17:09 [Docking@Home] Computer ID: 223; location: work; project prefs: work
2006-10-26 20:17:09 [SETI@home] Computer ID: 2185126; location: work; project prefs: default
2006-10-26 20:17:09 [rosetta@home] Computer ID: 211470; location: work; project prefs: default
2006-10-26 20:17:09 [---] General prefs: from SETI@home (last modified 2006-10-26 00:43:33)
2006-10-26 20:17:09 [---] General prefs: no separate prefs for work; using your defaults
2006-10-26 20:17:09 [---] Remote control not allowed; using loopback address
2006-10-26 20:17:09 [SETI@home] Deferring computation for result 15my03aa.5778.10817.823572.3.230_1
2006-10-26 20:17:09 [rosetta@home] Resuming computation for result 1hz6A_BOINC_NATIVEJUMPS_CLOSE_CHAINBREAKS_VARY_ALL_BOND_A$
Now, it's just a matter of waiting for it to get another WU. Because of the errors, it's only allowed to download 1 WU per day :-)
BTW, the machine has 2 GB of RAM and 2 GB of swap. Here's the output of free while it's running boinc and a Rosetta WU.
EDIT: If this doesn't work, I'm going to try increasing the max locked memory to 32, since that's what it is on my other linux box (FC3) that works. BTW, the RAM on the RHEL3 box is ECC and the machine works reliably with only a reboot for kernel or hardware upgrades and there haven't been any hardware upgrades in several months, maybe close to a year. Both boxes are running libsafe so that isn't it. Rosetta works great on both Linux boxes, as did Predictor@Home, before it stopped handing out work.
Have you tried with ulimit -s unlimited? I tested several stack limits, but only unlimited seems to go through. Let me know if that still gives you problems. If it does, we might be looking at another issue.
Thanks
Andre
OK, I changed /etc/init.d/boinc to set the stack to unlimited and I doubled the memory values to about 1 GB. Here's the boinc.log from where I started the service up again (as dumped from nano so long lines end in $).
core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 1024000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 4
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 1024000
2006-10-26 20:17:09 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-10-26 20:17:09 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-10-26 20:17:09 [---] Data directory: /home/BOINC
2006-10-26 20:17:09 [---] Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
2006-10-26 20:17:09 [---] Memory: 1.95 GB physical, 1.95 GB virtual
2006-10-26 20:17:09 [---] Disk: 16.02 GB total, 11.74 GB free
2006-10-26 20:17:09 [Docking@Home] Computer ID: 223; location: work; project prefs: work
2006-10-26 20:17:09 [SETI@home] Computer ID: 2185126; location: work; project prefs: default
2006-10-26 20:17:09 [rosetta@home] Computer ID: 211470; location: work; project prefs: default
2006-10-26 20:17:09 [---] General prefs: from SETI@home (last modified 2006-10-26 00:43:33)
2006-10-26 20:17:09 [---] General prefs: no separate prefs for work; using your defaults
2006-10-26 20:17:09 [---] Remote control not allowed; using loopback address
2006-10-26 20:17:09 [SETI@home] Deferring computation for result 15my03aa.5778.10817.823572.3.230_1
2006-10-26 20:17:09 [rosetta@home] Resuming computation for result 1hz6A_BOINC_NATIVEJUMPS_CLOSE_CHAINBREAKS_VARY_ALL_BOND_A$
Now, it's just a matter of waiting for it to get another WU. Because of the errors, it's only allowed to download 1 WU per day :-)
BTW, the machine has 2 GB of RAM and 2 GB of swap. Here's the output of free while it's running boinc and a Rosetta WU.
EDIT: If this doesn't work, I'm going to try increasing the max locked memory to 32, since that's what it is on my other linux box (FC3) that works. BTW, the RAM on the RHEL3 box is ECC and the machine works reliably with only a reboot for kernel or hardware upgrades and there haven't been any hardware upgrades in several months, maybe close to a year. Both boxes are running libsafe so that isn't it. Rosetta works great on both Linux boxes, as did Predictor@Home, before it stopped handing out work.
WTG David,
Looks to me as if it worked. To get a w/u from Docking, have you tryed suspeneding other experiments, then updating Docking? Should give you a w/u, or set the "preferences" at your account page to maybe 0.3 days? Worked for me , hope it does for you also.
Sincerely
Doug
OK, I changed /etc/init.d/boinc to set the stack to unlimited and I doubled the memory values to about 1 GB. Here's the boinc.log from where I started the service up again (as dumped from nano so long lines end in $).
So run_client is using a script in /etc/init.d as well.
I just tried to get this information for two days, by asking the content of that script (run_client). Changing that script is what I did to have my system working, as I said you can change what you want in /etc/security/limits.conf, but the changes will not be used by the client, 'cause those changes only apply to a shell instance. So, to have them working, you must start the client from a shell, i.e. using a script.
Thanks for your help Daniele. It's okay if the ulimit fix doesn't work for some people. We are working on another more permanent solution and when that is released, Charmm will hopefully work for all users.
Thanks for your help again! Much appreciated.
Andre
So, we finally came to something serious.
All my posts were headed to help docking@home, and help you at the same time.
But if you say you can run another project, this mean you're not a good alpha tester. In fact, you are not an alpha tester at all, since you can't manage your system.
I thoughy for a moment you wanted to help and to improve your know-how, but probably this is too diffcult for your, at this moment.
Have a good time crunching.
PS: docking@home IS working, fine
For sure it sometimes depends on the distro, but it's really rare. Also in this case, I was trying to have the content of run_client script, to analyze the operations and finally understand which other script it called.
This script is (as I now understand from other users' posts) in /etc/init.d and it's quite the same as debian's one. But I couldn't get those information by doug, so I get nervous 'cause I asked many times for the same thing and got no useful replies.
So, maybe it's because my difficulties in using a foreign languages in such a technical matter. For example I didn't understand if Doug starts the client by using the script (right move) or by directly clicking on the client executable file (wrong move).
Have you tried with ulimit -s unlimited? I tested several stack limits, but only unlimited seems to go through. Let me know if that still gives you problems. If it does, we might be looking at another issue.
Thanks
Andre
OK, I changed /etc/init.d/boinc to set the stack to unlimited and I doubled the memory values to about 1 GB. Here's the boinc.log from where I started the service up again (as dumped from nano so long lines end in $).
[code]core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 1024000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 4
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 1024000
You are on the right way, no doubt. I can't help you to choose the right values for your system, since it's a 2.4 kernel (so much different from 2.6, you know) and red hat management structure (different from debian).
For example my settings are far "larger" than yours (I have many unlimited resources by default) and everything is working fine, so in any case you shouldn't be afraid of raising the limits. Of course it would be better running the client as a daemon with a specific user (for example debian has the boinc user), so you can raise its limits safely, since that user will have only the permission to change its own files (boinc files). If you give unlimited permissions to your main user, this could be a problem with many other buggy applications.
OK, I changed /etc/init.d/boinc to set the stack to unlimited and I doubled the memory values to about 1 GB. Here's the boinc.log from where I started the service up again (as dumped from nano so long lines end in $).
So run_client is using a script in /etc/boinc as well.
I just tried to get this information for two days, by asking the content of that script (run_client). Changing that script is what I did to have my system working, as I said you can change what you want in /etc/security/limits.conf, but the changes will not be used by the client, 'cause those changes only apply to a shell instance. So, to have them working, you must start the client from a shell, i.e. using a script.
Anyway, have fun.
Daniele: I think you are confusing me (David) with Doug. I have boinc running as a service on a non-graphical Redhat Enterprise Linux 3 machine. Everything on that machine is done via command line. Boinc is running as a daemon. On that version of Linux, /etc/security/limits.conf is ignored unless you re-configure PAM. I tried that earlier and set the stack limit for user boinc to a large number, but the Docking@Home work unit still got the error, so I backed out the PAM changes. I suspect that, in addition to the unlimited stack change, I am also going to have to set the max locked memory higher by doing "ulimit -l 32". By the way, even on my other Linux machine which has a graphical interface, I do almost everything on Linux from a command line. I am 50 years old and started programing during the 1970's, in the days of CP/M on the 8080/8085/Z-80, OS/8 on the PDP-8, timeshare terminals on CDC (Control Data Corporation) mainframes, and I even did assembler and Cobol using punched cards on IBM 360 and 370. I have even thought about writing a command line boinc manager (similar to top) that uses curses for those of us who have some Linux machines that are command line only.
Doug: The problem is not that it does not ask for work. The one Work Unit a day limit is imposed by the Docking@Home server because that machine always returns bad results. It is already set to update every 0.2 days.
Daniele: I think you are confusing me (David) with Doug. I have boinc running as a service on a non-graphical Redhat Enterprise Linux 3 machine. Everything on that machine is done via command line. Boinc is running as a daemon. On that version of Linux, /etc/security/limits.conf is ignored unless you re-configure PAM. I tried that earlier and set the stack limit for user boinc to a large number, but the Docking@Home work unit still got the error, so I backed out the PAM changes. I suspect that, in addition to the unlimited stack change, I am also going to have to set the max locked memory higher by doing "ulimit -l 32". By the way, even on my other Linux machine which has a graphical interface, I do almost everything on Linux from a command line. I am 50 years old and started programing during the 1970's, in the days of CP/M on the 8080/8085/Z-80, OS/8 on the PDP-8, timeshare terminals on CDC (Control Data Corporation) mainframes, and I even did assembler and Cobol using punched cards on IBM 360 and 370. I have even thought about writing a command line boinc manager (similar to top) that uses curses for those of us who have some Linux machines that are command line only.
:D I'm not confusing you with doug, I "used" your words to imply something on what I was trying to know from Doug :)
If you are using the "standard" boinc package (as doug), i.e not the debian's one, you probably have the same files as Doug. Anyway, it doesn't matter anymore (at least for now).
Regarding limits.conf, I'll repeat what I've learned. But first I have some questions, since I'm not sure I'm using the right words to say what I want to, and I'n not sure I understand every single word of yours.
Given that:
you're using kernel 2.4 (different from 2.6)
you're using Red Hat (different from Debian)
then I'd like to make you a question or two, just to understand if I have understood.
First of all, are you using the package with run_client script or another package (mine has not, for example)?
You changed PAM settings and modified limits.conf, then got the same errors.
Did I understand?
When you start the core_client, which user do you use and with wich user's permissions do you run the client? For example, I run the client by starting the script as root or daniele, but the client runs with boinc user's permissions (I do this by using the `start-stop daemon` command with the --user parameter set to $BOINC_USER, so I can change the "running user" by modifying that variable).
I'm saying this because if you change, let's make an example. If you modify limits.conf by setting unlimites stack for "David" user, but you start the script with root, the changes won't be used because those limits apply only to a shell instance eventually started by "David", not to a generic process owned by "David". So the changes will work only if you start the client using "David", if the client runs with David permissions.
I had a similar problem modifying limits.conf for "boinc" user and not having that changes working, since the client was started by root at boot time.
Also if you think I can't help you, would you mind answering just to let me learn? :)
I suspect that, in addition to the unlimited stack change, I am also going to have to set the max locked memory higher by doing "ulimit -l 32".
Yes, as I wrote in one post, you probably have too strict limits, I wrote that I have much more unlimited resources than you, so it could be your problem is only from that.
I am 50 years old and started programing during the 1970's, in the days of CP/M on the 8080/8085/Z-80, OS/8 on the PDP-8, timeshare terminals on CDC (Control Data Corporation) mainframes, and I even did assembler and Cobol using punched cards on IBM 360 and 370. I have even thought about writing a command line boinc manager (similar to top) that uses curses for those of us who have some Linux machines that are command line only.
I'm fond of experienced programmers, I'd have a whole bunch of questions for you. I'm still young, anyway I've my little experience too, being an informatics engineer, though I'm not so expert in linux administration, I study computational linguistics (and I've to improve my english for this).
When I finish my current work, I'd like to set up a boinc project for natural language processing ENTIRELY based on neural networks. I really hate neural networks, but they permit the distribution of processing.
I am 50 years old and started programing during the 1970's, in the days of CP/M on the 8080/8085/Z-80, OS/8 on the PDP-8, timeshare terminals on CDC (Control Data Corporation) mainframes, and I even did assembler and Cobol using punched cards on IBM 360 and 370. I have even thought about writing a command line boinc manager (similar to top) that uses curses for those of us who have some Linux machines that are command line only.
I'm fond of experienced programmers, I'd have a whole bunch of questions for you. I'm still young, anyway I've my little experience too, being an informatics engineer, though I'm not so expert in linux administration, I study computational linguistics (and I've to improve my english for this).
When I finish my current work, I'd like to set up a boinc project for natural language processing ENTIRELY based on neural networks. I really hate neural networks, but they permit the distribution of processing.
Dear Dannielle,
I apologise for not understanding what you would like me too do.I am dissapointed in "self" and feel less than for not being able to get this to work.Honestly Danniele, I am a computer Dummy.3 years on a P.c., and 2 running linux by "guessing" Woops, Just want you too know I want to fix this problem.
But donnot understand many things about a P.C.,I think it is commendable for you to be so helpful, to start your own Boinc project also.What is the thing {for lack of better knowledge} that you would like me too try?Posting seems so impersonal, but, would like to try again.Fully understand if you donnot.Be Well,
and Good luck with your future indevours.
Dave, thnks again for your help.Really expected when we reset the stacksize, it would work.I think Dannielle is right also that I donnot know my own Distro.This has lead me to be unsure of self and am not running Boinc 24-7 any more, for fear that I installed Boinc incorrectly the last 3 installs.Hope you can get up and running for Docking, and hope I can get my Distro to not error out also.35
days of trying has left me dissapointed and Angry.Myself am not running Boinc as a Daemen, donnot know what a Daeman is? And yes I have read the Manual, as vague
as it is, and donnot understand a few things.My girlfriend is from Southern Italy, know a little "love" talk in Italian, it is commendable and great that you know 2 Languages, have enough difficulties with English.I donnot know P.C.
terminology at all.Sat down at a P.C. at the age of 45, 48 now and not much has changed, still a dummy.{Smile}
Be Well Dannielle and David
Sincerely
"Happy Crunchung"
Doug
i686 32bit Linux 2.6.16.27.tex1 kernel (4GB highmem)
The kernel package contains the Linux 2.6.16.27.tex1 kernel, the core of your
PCLinux operating system. It is optimised for the i686 cpu arch and supports up to
and including 4GB of system memory (via highmem).
Dave, or Dannielle
Would a new Kernel maybe help?I have many choices to use from my Repos, above is one choice.
Thanks
Doug
I apologise for not understanding what you would like me too do
Don't be sorry Doug, the only important thing is if you want to try another time. :) I understand if you prefer to save your time and crunch for another project.
First of all, let's collect useful informations, so we can understand if it's a problem in your distro (which I can't solve now, I'm sorry) or a little solvable one.
To follow a clean path, maybe I'll need to try things that you have tried yet.
Open a console ("a shell") and type (note, every row is followed by RETURN key)
cd /home/doug/BOINC
ls -al
this will output a list of files in that directory, I need that list (copy and paste it here).
Then type
cd /etc/init.d
ls -al *boinc*
and post here the output.
Then type
ulimit -a
and post the output.
This would be a good start, later I'll tell you what to do with certain files.
Bye doug, I hope we can make docking work on your pc :D
i686 32bit Linux 2.6.16.27.tex1 kernel (4GB highmem)
The kernel package contains the Linux 2.6.16.27.tex1 kernel, the core of your
PCLinux operating system. It is optimised for the i686 cpu arch and supports up to
and including 4GB of system memory (via highmem).
Dave, or Dannielle
Would a new Kernel maybe help?I have many choices to use from my Repos, above is one choice.
Thanks
Doug
I'll try to explain in simple words, also because I don't know complicated words :)
The problem is not the kernel itself (2.6.x is good and usually up to date), but how the distro is done inside. Every distro has its own way to manage the system and when one tries to create a distro and to use new ways to manage it, often the result is messy (to create a good distro is very difficult).
It happened and happens to distros derived from Red Hat and Mandriva (and also Suse) that the internal structure of the whole distro lacks something, or contains important errors. It's difficult for me to explain, but we could say that most of these distros are incoherent in their structures. And the problem gets worse and worse since they permit to install "incoherent" packages.
The basic idea is good, to have the most general structure and install whatever package you want, but in the real world this doesn't work well The dimonstration is that debian has the only real working structure (there's slackware, of course, but it's a different matter), having a very well engineered internal structure, very very coherent and rigid.
How can we guess the right kernel package if it happens that the distro lacks a coherent managing structure? It could happen that an application works well with certain configurations, and not with other ones. So, let's proceed step by step, trying to understand if this is a docking's issue or your distro's, or (it's possible) you only need to adjust something in the way you operate your system :)
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd/doug/home/BOINC
bash: cd/doug/home/BOINC: No such file or directory
bash-3.00$ ls-al
bash: ls-al: command not found
bash-3.00$ ls - al
ls: -: No such file or directory
ls: al: No such file or directory
bash-3.00$ cd/etc./init.d
bash: cd/etc./init.d: No such file or directory
bash-3.00$ cd/etc/init.d
bash: cd/etc/init.d: No such file or directory
bash-3.00$
Dear Dannielle,
Thanks for your ongoing help.Must be me and this Distro that is going wrong.
But will do all that is neccasary to help you, help me and Docking.First, I used
the Linux console and got this above.At the top doug@localhost:command not found
was there when I trped this,Will clear it next time, if possible.I am not getting "output" from the commands.No such file, yet I see the file in Home, just as you said.Thanks for letting me know too enter after each line, help for dummies.Have those commands stored on Kedit and will keep trying for some results for you.
Thanks, will try again
ash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd/home/doug/BOINC
bash: cd/home/doug/BOINC: No such file or directory
bash-3.00$ ls-al
bash: ls-al: command not found
bash-3.00$ cd/etc/init.d
bash: cd/etc/init.d: No such file or directory
bash-3.00$ ulimit -s
unlimited
bash-3.00$
Tried again Dannielle,
NOTE, the ulimit -s command gave me an unlimited from saving it as so.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd/doug/home/BOINC
bash: cd/doug/home/BOINC: No such file or directory
bash-3.00$ ls-al
bash: ls-al: command not found
bash-3.00$ ls - al
ls: -: No such file or directory
ls: al: No such file or directory
bash-3.00$ cd/etc./init.d
bash: cd/etc./init.d: No such file or directory
bash-3.00$ cd/etc/init.d
bash: cd/etc/init.d: No such file or directory
bash-3.00$
Ok Doug, here there are few things for you to learn.
You wrote this
bash-3.00$ cd/doug/home/BOINC
but there are 2 mistakes.
The first is that between cd (the command) and the directory where we want to go (the argument of the command) we need a space (blank).
So
cd <path_to_the_directory>
Then, the directory is not
/doug/home/BOINC
This directory doesn't exist.
The right one is
/home/doug/BOINC
So, the complete command + argument is
cd /home/doug/BOINC
Then
ls -al
remember the SPACE between the command ls and the arguments -al
If you prefer, I can send you a script (a little shell program) by mail to collect some informations :)
You'll only have to copy it on your desktop and type a line in the console.
bash-3.00$ ls -al
total 8472
drwxr-xr-x 3 doug doug 4096 Oct 26 08:34 .
drwxr-xr-x 38 doug doug 4096 Oct 27 18:55 ..
-rw-r--r-- 1 doug doug 256 May 5 15:36 binstall.sh
-rw-r--r-- 1 doug doug 257 Sep 25 15:09 binstall.sh~
-rwxr-xr-x 1 doug doug 1744688 May 5 15:36 boinc
-rwxr-xr-x 1 doug doug 434320 May 5 15:36 boinc_cmd
-rwxr-xr-x 1 doug doug 6402756 May 5 15:36 boincmgr
-rw-r--r-- 1 doug doug 1264 May 5 15:36 boincmgr.16x16.png
-rw-r--r-- 1 doug doug 1786 May 5 15:36 boincmgr.32x32.png
-rw-r--r-- 1 doug doug 558 May 5 15:36 boincmgr.8x8.png
-rw-r--r-- 1 doug doug 1857 Oct 26 08:34 client_state.xml
-rw-r--r-- 1 doug doug 1857 Oct 26 08:34 client_state_prev.xml
-rw-r--r-- 1 doug doug 0 Oct 26 08:34 do_fp
-rw------- 1 doug doug 32 Aug 25 06:20 gui_rpc_auth.cfg
drwxr-xr-x 33 doug doug 4096 May 5 15:36 locale
-rw-r--r-- 1 doug doug 0 Aug 25 06:20 lockfile
-rwxr-xr-x 1 doug doug 41 Sep 26 11:54 run_client
-rwxr-xr-x 1 doug doug 44 Oct 11 15:20 run_manager
-rwxr-xr-x 1 doug doug 44 Oct 11 15:20 run_manager~
-rw-r--r-- 1 doug doug 0 Aug 25 06:20 stderrdae.txt
-rw-r--r-- 1 doug doug 5281 Oct 26 08:34 stdoutdae.txt
bash-3.00$
Hello Danniele,
Looks like you are a GOOD Teacher, at least I got a print from the command, Wow
those are read write in the files, no that much. anyways.I have No idea what to do, but, hope this is of help to you.If you want to send a thing by mail, would you like my gmail address?Will the information "be of service to you"?
Hi Doug :)
Only a question or two.
Have you reinstalled the boinc package recently or this is an old installation?
Anyway, now we'll read the content of the script named run_client
when you are in this directory, by typing
cd /home/doug/BOINC
we give the command
cat run_client
copy and paste the output here.
If you prefer, you can open it in kedit and copy and paste.
Did you modify the run_manager script? And why not the run_client_script?
But the real question is: have you ever used run_client to start the client, or you only did it by clicking the link in the kde menu?
Regarding the mail, if this problem needs particularly difficult commands maybe it's better if I send you a script. Let's see what happens.
I need also the output of this command.
While the boinc system (the client, the manager and one app, not Charmm for now) is running, type
ps axu | grep boinc
This is important and we'll need it to see something about the client and Charmm while they are running. Maybe later we'll add some options to know something particular, if necessary.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ cat run_client
cd "/home/doug/BOINC" && exec ./boinc $@
bash-3.00$
I am sorry Dannielle, I used the shell entered cd /home/doug/BOINC "Enter"
than I typed cat run_client.
I have never started Boinc with with run_client, only KDE Menu
first I click on the BOINC, and then, I click on the Boinc Manager
also in the KDE Menu
After my Manager got that error, as seen on {screen shots}, I overwrote the installed Boinc from 2 Months prior {when I applyed this system 2 months ago}
about 9 days ago I overwrote BOINC, I downloaded a new shell script and put it over top of the old one.
Did you want me to go right into run_client with Kedit? I must have done this wrong also, still affected from Medication, awful stuff and Morphine, apologise
for my daftness{Dumbness} I will ive NO-Biggy
Let me know what is best
Thank you
Doug
Just saw your last post,
How do I type "Aproximitly" ps axu [ grep home, what is in the middle, cannot see it, too confirm it, also, is this done with a shell, or Kedit and in run_manager?, or with the "Shell" program?
Hello Dannielle,
It is Friday night at 9:30, must go to sleep.Will be "On Line" again in 10
Hours from now, around 8A.M. Saturday.I donnot know the time in Italy, but
just wanted you to know I will be back in 10 Hours from now.Thank-you
How do I type "Aproximitly" ps axu [ grep home, what is in the middle, cannot see it, too confirm it, also, is this done with a shell, or Kedit and in run_manager?, or with the "Shell" program?
It's the pipe |
Usually it's on the left of key 1, just above the symbol \
Hello Dannielle,
It is Friday night at 9:30, must go to sleep.Will be "On Line" again in 10
Hours from now, around 8A.M. Saturday.I donnot know the time in Italy, but
just wanted you to know I will be back in 10 Hours from now.Thank-you
Sincerely
Doug*
Here it is Saturday, 3:30 am :)
Have a good night, I hope with good news for tomorrow :)
:D I'm not confusing you with doug, I "used" your words to imply something on what I was trying to know from Doug :)
I'm sorry for the confusion on my part!
If you are using the "standard" boinc package (as doug), i.e not the debian's one, you probably have the same files as Doug. Anyway, it doesn't matter anymore (at least for now).
Regarding limits.conf, I'll repeat what I've learned. But first I have some questions, since I'm not sure I'm using the right words to say what I want to, and I'n not sure I understand every single word of yours.
Given that:
you're using kernel 2.4 (different from 2.6)
you're using Red Hat (different from Debian)
That is correct.
then I'd like to make you a question or two, just to understand if I have understood.
First of all, are you using the package with run_client script or another package (mine has not, for example)?
I downloaded the standard 5.2.13 BOINC client, but I am NOT using run_client or run_manager. The machine boots into text mode on the console. It is not setup to allow graphics mode. I control it by using ssh from my Windows XP machine.
You changed PAM settings and modified limits.conf, then got the same errors.
Did I understand?
Yes, that is correct.
When you start the core_client, which user do you use and with wich user's permissions do you run the client? For example, I run the client by starting the script as root or daniele, but the client runs with boinc user's permissions (I do this by using the `start-stop daemon` command with the --user parameter set to $BOINC_USER, so I can change the "running user" by modifying that variable).
I'm saying this because if you change, let's make an example. If you modify limits.conf by setting unlimites stack for "David" user, but you start the script with root, the changes won't be used because those limits apply only to a shell instance eventually started by "David", not to a generic process owned by "David". So the changes will work only if you start the client using "David", if the client runs with David permissions.
I had a similar problem modifying limits.conf for "boinc" user and not having that changes working, since the client was started by root at boot time.
Also if you think I can't help you, would you mind answering just to let me learn? :)
You can probably help me, and it is interesting to see how different Linux distributions do things. BOINC runs as a daemon and is started at boot. The command on RedHat to manually control a daemon is
service daemon_name {start|stop|restart|status}
I have it configured to start at boot time. If you do a "chkconfig --list", there is a line in the output which says
boinc 0:off 1:off 2:off 3:on 4:on 5:on 6:off
I'm using a modified version of the script which was pointed to by the BOINC website for running BOINC as a service. You can find a copy of my current modified script at
http://www.booksnbytes.com/boinc.txt
Fixing the stack, either through limits.conf or the init script doesn't seem to fix the problem so I've added "ulimit -l 32" as well as the other modifications. The default, even for root is "ulimit -l 4" on this release.
I added another command in front of the one which actually starts boinc that dumps the results of "ulimit -a" to the boinc log. The 2 lines that actually start boinc running are
su $BOINCUSER -c "ulimit -a" >>$LOGFILE 2>>$ERRORLOG
su $BOINCUSER -c "nice --adjustment=19 $BOINCEXE $BOINCOPTS" >>$LOGFILE 2>>$ERRORLOG &
Just in case it is different for Debian, su is the command to run a shell with substitute user and group IDs.
They made some changes to the command line options for boinc in 5.4.9 for Linux, so I will need to modify the script again for that version. That's why I'm still on BOINC 5.2.13 . I don't want to update the BOINC version until I get Docking@Home stabilized.
Here's the beginning of the boinc log file which shows the output of the "ulimit -a" and the output from BOINC as it starts up.
core file size (blocks, -c) 1000000
data seg size (kbytes, -d) 1536000
file size (blocks, -f) unlimited
max locked memory (kbytes, -l) 32
max memory size (kbytes, -m) unlimited
open files (-n) 4096
pipe size (512 bytes, -p) 8
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) 14335
virtual memory (kbytes, -v) 1536000
2006-10-27 14:32:48 [---] Starting BOINC client version 5.2.13 for i686-pc-linux-gnu
2006-10-27 14:32:48 [---] libcurl/7.14.0 OpenSSL/0.9.8 zlib/1.2.3
2006-10-27 14:32:48 [---] Data directory: /home/BOINC
2006-10-27 14:32:49 [---] Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
2006-10-27 14:32:49 [---] Memory: 1.95 GB physical, 1.95 GB virtual
2006-10-27 14:32:49 [---] Disk: 16.02 GB total, 11.73 GB free
2006-10-27 14:32:49 [Docking@Home] Computer ID: 223; location: work; project prefs: work
2006-10-27 14:32:49 [SETI@home] Computer ID: 2185126; location: work; project prefs: default
2006-10-27 14:32:49 [rosetta@home] Computer ID: 211470; location: work; project prefs: default
2006-10-27 14:32:49 [---] General prefs: from SETI@home (last modified 2006-10-26 00:43:33)
2006-10-27 14:32:49 [---] General prefs: no separate prefs for work; using your defaults
2006-10-27 14:32:49 [---] Remote control not allowed; using loopback address
2006-10-27 14:32:49 [SETI@home] Resuming computation for result 15my03aa.5778.10817.823572.3.230_1 using setiathome_enhanced$
2006-10-27 14:32:49 [rosetta@home] Deferring computation for result 1hz6A_BOINC_NATIVEJUMPS_SAVE_ALL_OUT__1306_34680_0
2006-10-27 15:05:21 [---] request_reschedule_cpus: project op
2006-10-27 15:05:22 [SETI@home] Pausing result 15my03aa.5778.10817.823572.3.230_1 (left in memory)
2006-10-27 15:05:22 [rosetta@home] Restarting result 1hz6A_BOINC_NATIVEJUMPS_SAVE_ALL_OUT__1306_34680_0 using rosetta versio$
2006-10-27 15:05:26 [Docking@Home] Sending scheduler request to http://docking.utep.edu/docking_cgi/cgi
2006-10-27 15:05:26 [Docking@Home] Reason: Requested by user
2006-10-27 15:05:26 [Docking@Home] Requesting 17280 seconds of new work
2006-10-27 15:05:31 [Docking@Home] Scheduler request to http://docking.utep.edu/docking_cgi/cgi succeeded
2006-10-27 15:05:31 [Docking@Home] Message from server: No work sent
2006-10-27 15:05:31 [Docking@Home] Message from server: (reached daily quota of 1 results)
2006-10-27 15:10:39 [---] request_reschedule_cpus: project op
Since you asked Doug for it, I also included the output of "ps axu|grep boinc" while it is running, but nano (text editor similar to an extended version of pico) cut off the end of some of the long lines of options to Rosetta. It's not possible to catch a Docking@Home work unit because they only last about 700 seconds and the machine has returned so many errors on work units that it's only allowed to download 1 work unit a day by the Docking@Home BOINC server software.
root 7835 0.0 0.0 5192 812 ? S Oct27 0:00 su boinc -c nice --adjustment=19 /home/BOINC/boinc -return_re$
boinc 7836 0.0 0.1 4688 2512 ? SN Oct27 0:00 /home/BOINC/boinc -return_results_immediately
boinc 7840 0.0 1.4 38112 29360 ? SN Oct27 0:00 setiathome-5.12.i686-pc-linux-gnu
boinc 7841 0.0 1.4 38112 29360 ? SN Oct27 0:00 setiathome-5.12.i686-pc-linux-gnu
boinc 7842 15.0 1.4 38112 29360 ? SN Oct27 87:36 setiathome-5.12.i686-pc-linux-gnu
boinc 7843 0.0 1.4 38112 29360 ? SN Oct27 0:00 setiathome-5.12.i686-pc-linux-gnu
boinc 11641 92.4 3.1 85656 63568 ? RN Oct27 201:15 rosetta_5.34_i686-pc-linux-gnu xx 1dtj _ -output_silent_gz -s$
boinc 11649 0.0 3.1 85656 63568 ? SN Oct27 0:00 rosetta_5.34_i686-pc-linux-gnu xx 1dtj _ -output_silent_gz -s$
boinc 11652 0.0 3.1 85656 63568 ? SN Oct27 0:00 rosetta_5.34_i686-pc-linux-gnu xx 1dtj _ -output_silent_gz -s$
boinc 11656 0.0 3.1 85656 63568 ? SN Oct27 0:00 rosetta_5.34_i686-pc-linux-gnu xx 1dtj _ -output_silent_gz -s$
I'll probably have to edit this to make sure the BBcodes I used are balanced properly :-)
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3054 0.0 0.3 4472 2424 ? S 05:19 0:00 /home/doug/BOINC/boinc
doug 3058 0.3 0.7 10040 6076 ? S 05:19 0:01 /home/doug/BOINC/boincmgr
doug 3118 0.0 0.0 1784 484 pts/0 S 05:24 0:00 grep boinc
bash-3.00$
This is "all" I received from ps axu | grep boinc, while it is running, Malaria Control, also addded the line using Kedit "ulimit -s unlimited" in run_manger
Sat Oct 28 05:19:19 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sat Oct 28 05:19:19 2006||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
Sat Oct 28 05:19:19 2006||Data directory: /home/doug
Sat Oct 28 05:19:19 2006||Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
Sat Oct 28 05:19:19 2006||Memory: 749.28 MB physical, 1.56 GB virtual
Sat Oct 28 05:19:19 2006||Disk: 18.62 GB total, 17.68 GB free
Sat Oct 28 05:19:19 2006|climateprediction.net|URL: http://climateprediction.net/; Computer ID: 473373; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|malariacontrol.net beta|URL: http://www.malariacontrol.net/; Computer ID: 7568; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|Docking@Home|URL: http://docking.utep.edu/; Computer ID: 98; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|boincsimap|URL: http://boinc.bio.wzw.tum.de/boincsimap/; Computer ID: 37849; location: home; project prefs: default
Sat Oct 28 05:19:19 2006||General prefs: from malariacontrol.net beta (last modified 2006-10-25 17:15:35)
Sat Oct 28 05:19:19 2006||General prefs: no separate prefs for home; using your defaults
Sat Oct 28 05:19:19 2006||Local control only allowed
Sat Oct 28 05:19:19 2006||Listening on port 31416
Sat Oct 28 05:19:19 2006|climateprediction.net|Deferring task hadcm3lbm_blbb_25310106_0
Sat Oct 28 05:19:19 2006|malariacontrol.net beta|Resuming task wu_11_34_10204_0_2120349614_0 using malariacontrol version 542
Sat Oct 28 05:19:19 2006||Using earliest-deadline-first scheduling because computer is overcommitted.
Sat Oct 28 05:19:19 2006||Suspending work fetch because computer is overcommitted.
All I have in my que is a climate Model and a Malaria w/u, yet P.C. is overcommited
*
Builds and runs under KDE, QT3-only on Linux, Windows and Mac OS/X.
Despite comments to the contrary on a certain other website, Rekall is
first and foremost a KDE application; non-KDE builds are supported by a
compatability library. Builds as a full and a run-time version; the
run-time version has no design functionality.
*
Database documents (forms, reports, etc.) can be stored in either the
database or the filesystem. Database allows easy sharing, etc., file
system can be used if (for example) you have read-only access to the
database and just want to view it.
Just came in, ran a charm w/u just for fun, errored out, but, got further this time.4.54%
Sincerely
Doug
I'll probably have to edit this to make sure the BBcodes I used are balanced properly :-)
As I said before, could be only a simple problem of permissions, you have to wait until a WU is sent. Now I can see the user who runs the client is boinc, and your script is obviously not started by boinc user (I suppose it's done by root, or by your main user). This way you have to set the permission (for example ulimit -s unlimited, or whatever you want) for the user who starts the script, not for boinc client.
In my case, boinc user has ulimit -s 8192, but when root starts the script has (for a little time) ulimit -s unlimited. So it starts a shell with unlimited stack, this shell creates the daemon with unlimited stack and change the uid of the daemon in boinc's uid. Then the shell dies, the client daemon becomes session manager, and it remains with unlimited stack. But, I repeat, boinc user has not particular permissions :)
Try this, and increase the limits (yours are too strict).
Here's the output of `ulimit -a` when I start the client
daniele@lele:~$ ulimit -a
core file size (blocks, -c) 0
data seg size (kbytes, -d) unlimited
max nice (-e) 20
file size (blocks, -f) unlimited
pending signals (-i) unlimited
max locked memory (kbytes, -l) unlimited
max memory size (kbytes, -m) unlimited
open files (-n) 1024
pipe size (512 bytes, -p) 8
POSIX message queues (bytes, -q) unlimited
max rt priority (-r) unlimited
stack size (kbytes, -s) unlimited
cpu time (seconds, -t) unlimited
max user processes (-u) unlimited
virtual memory (kbytes, -v) unlimited
file locks (-x) unlimited
Probably debian mantainers choosen such high limits because the debian structure is very coherent and stable, so it's not a problem (in fact it's not).
Sorry if I can't stay more time on the problem, I've a very difficult exam in few days!
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3054 0.0 0.3 4472 2424 ? S 05:19 0:00 /home/doug/BOINC/boinc
doug 3058 0.3 0.7 10040 6076 ? S 05:19 0:01 /home/doug/BOINC/boincmgr
doug 3118 0.0 0.0 1784 484 pts/0 S 05:24 0:00 grep boinc
bash-3.00$
This is "all" I received from ps axu | grep boinc, while it is running, Malaria Control, also addded the line using Kedit "ulimit -s unlimited" in run_manger
Hi Doug :)
Maybe it's a mistake in writing, but you have to add "ulimit -s unlimited" as the first line in run_client, not run_manager (the manager doesn't need such a permission).
So, modify run_client script, then save it.
Then stop the client, and since I don't know how your client works, it's better if you restart the whole system.
When you have restarted, don't start the client in kde, but open a console and type
cd /home/doug/BOINC
./run_client -redirectio -dir /home/doug/BOINC
Note: not REDIRECTION, but REDIRECTIO
So you will use the script to start the client, and the unlimited stack should be used.
Then launch the manager the way you usually do it.
After launching the client
ps axu | grep boinc
and post the output.
This is the simplest thing we can do, a starting point.
PS: it's probable that you have en error in .bashrc or .bash_profile, those lines on the top of the shell (8192: command not found etc) are not good.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3054 0.0 0.3 4472 2424 ? S 05:19 0:00 /home/doug/BOINC/boinc
doug 3058 0.3 0.7 10040 6076 ? S 05:19 0:01 /home/doug/BOINC/boincmgr
doug 3118 0.0 0.0 1784 484 pts/0 S 05:24 0:00 grep boinc
bash-3.00$
This is "all" I received from ps axu | grep boinc, while it is running, Malaria Control, also addded the line using Kedit "ulimit -s unlimited" in run_manger
Hi Doug :)
Maybe it's a mistake in writing, but you have to add "ulimit -s unlimited" as the first line in run_client, not run_manager (the manager doesn't need such a permission).
So, modify run_client script, then save it.
Then stop the client, and since I don't know how your client works, it's better if you restart the whole system.
When you have restarted, don't start the client in kde, but open a console and type
cd /home/doug/BOINC
./run_client -redirectio -dir /home/doug/BOINC
Note: not REDIRECTION, but REDIRECTIO
So you will use the script to start the client, and the unlimited stack should be used.
Then launch the manager the way you usually do it.
After launching the client
ps axu | grep boinc
and post the output.
This is the simplest thing we can do, a starting point.
PS: it's probable that you have en error in .bashrc or .bash_profile, those lines on the top of the shell (8192: command not found etc) are not good.
Hello Dannielle,
yes, that was a typing error, I saved the ulimit -s unlimited in the correct file.When I did the commands, received no OUTPUT, but, would not let me start
the Manager due to "wrong password", LOL This is fun actually, I understand
your School is more Important than helping a dummy, Was able to get Boinc going the old way, seeing as this way did not, may have mistyped, so will continue typing and trying.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 6378 0.0 0.3 4472 2432 ? S 10:20 0:00 /home/doug/BOINC/boinc
doug 6386 0.2 0.7 10040 6072 ? S 10:20 0:00 /home/doug/BOINC/boincmgr
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ./run_client -redirectio -dir /home/doug/BOINC
Has to be the localhost not found, BIG problem ehh?
Hello Dannielle,
yes, that was a typing error, I saved the ulimit -s unlimited in the correct file.When I did the commands, received no OUTPUT, but, would not let me start
the Manager due to "wrong password", LOL This is fun actually, I understand
your School is more Important than helping a dummy, Was able to get Boinc going the old way, seeing as this way did not, may have mistyped, so will continue typing and trying.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 6378 0.0 0.3 4472 2432 ? S 10:20 0:00 /home/doug/BOINC/boinc
doug 6386 0.2 0.7 10040 6072 ? S 10:20 0:00 /home/doug/BOINC/boincmgr
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ./run_client -redirectio -dir /home/doug/BOINC
Has to be the localhost not found, BIG problem ehh?
So, what happened exactly?
You restarted the system, then typed
./run_client -redirectio -dir /home/doug/BOINC
and the manager didn't connect to the client?
This could be due to how the manager works, I don't know it very well yet.
Anyway, if things are this way, when you restart the system (everytime you start the client you have to stop it first, and since it's difficult for you to safely kill the process, maybe it's better if you reboot the system) try this. The problem of manager not connecting to the client is not a real problem, we only need to learn or (better) someone who knows about this stuff.
Try this, and let's see the output (remember, don't use the usual way to start the client and the manager, use only the console).
I need a `ps axu | grep boinc` BEFORE executing any other application
Hello Dannielle,
yes, that was a typing error, I saved the ulimit -s unlimited in the correct file.When I did the commands, received no OUTPUT, but, would not let me start
the Manager due to "wrong password", LOL This is fun actually, I understand
your School is more Important than helping a dummy, Was able to get Boinc going the old way, seeing as this way did not, may have mistyped, so will continue typing and trying.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 6378 0.0 0.3 4472 2432 ? S 10:20 0:00 /home/doug/BOINC/boinc
doug 6386 0.2 0.7 10040 6072 ? S 10:20 0:00 /home/doug/BOINC/boincmgr
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ./run_client -redirectio -dir /home/doug/BOINC
Has to be the localhost not found, BIG problem ehh?
So, what happened exactly?
You restarted the system, then typed
./run_client -redirectio -dir /home/doug/BOINC
and the manager didn't connect to the client?
This could be due to how the manager works, I don't know it very well yet.
Anyway, if things are this way, when you restart the system (everytime you start the client you have to stop it first, and since it's difficult for you to safely kill the process, maybe it's better if you reboot the system) try this. The problem of manager not connecting to the client is not a real problem, we only need to learn or (better) someone who knows about this stuff.
Try this, and let's see the output (remember, don't use the usual way to start the client and the manager, use only the console).
I need a `ps axu | grep boinc` BEFORE executing any other application
Hello Dannielle,
what happened, Boinc did not start up, checked thru KSGaurd, then tried starting the Boinc the old way and said I was using the wrong password? Go Figure ?
This is the OUTPUT from the last try
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3076 0.0 0.0 1788 488 pts/0 S 11:10 0:00 grep boinc
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ps axu | grep boinc
doug 3078 0.0 0.0 1752 244 pts/0 R 11:12 0:00 grep boinc
bash-3.00$ ./run_client -redirectio -dir /home/doug BOINC
Unknown option: BOINC
Usage: ./boinc [options]
-help show options
-version show version info
-exit_when_idle Get/process/report work, then exit
-show_projects show attached projects
-return_results_immediately contact server when have results
-detach_project <URL> detach from a project
-reset_project <URL> reset (clear) a project
-attach_project <URL> <key> attach to a project
-update_prefs <URL> contact a project to update preferences
-run_cpu_benchmarks run the CPU benchmarks
-check_all_logins for idle detection, check remote logins
too -allow_remote_gui_rpc allow remote GUI RPC connections
-gui_rpc_port port for GUI RPCs
-redirectio redirect stdout and stderr to log files
-detach detach from console (Windows)
-dir <path> use given dir as BOINC home
-no_gui_rpc don't allow GUI RPC, don't make socket
-daemon run as daemon (Unix)
bash-3.00$ ./ run_manager --verbose
bash: ./: is a directory
bash-3.00$ ps axu | grep boinc
doug 3082 0.0 0.0 1788 480 pts/0 R 11:14 0:00 grep boinc
bash-3.00$
Sorry for the poor results, just trying "My Best"{nice to see run as Daemon}
Sincerely
Doug
I read something about manager not being able to connect, and it's really a dumb problem (it seems like a little bug or similar in the manager).
Anyway, it's solvable :)
After the sequence I wrote before, type
ls -al (just to let me see which new files have been created)
cat gui_rpc_auth.cfg
and let me see what's in.
It seems that the manager is not able to remember blank passwords :)
For now don't make what follows...
If this is your problem, then you'll need to rewrite the password in that file (choose one, but be careful, the file must contain only the password, with a RETURN character at the end), and if it works you will return to the blank password (this way it will remember it, a real dumb problem).
It the password in gui_rpc_auth.cfg is not blank, then use that password to connect to the client. If it works, then we'll modify the password to the blank one.
Hello Dannielle,
what happened, Boinc did not start up, checked thru KSGaurd, then tried starting the Boinc the old way and said I was using the wrong password? Go Figure ?
This is the OUTPUT from the last try
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3076 0.0 0.0 1788 488 pts/0 S 11:10 0:00 grep boinc
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ps axu | grep boinc
doug 3078 0.0 0.0 1752 244 pts/0 R 11:12 0:00 grep boinc
bash-3.00$ ./run_client -redirectio -dir /home/doug BOINC
Unknown option: BOINC
Yes, now I'm on my way to figure out how the manager works :)
Anyway, you made a mistake typing the command above, it lacks the / between doug and BOINC.
Note, it's a fundamental step that you reboot the system before restarting the client (when we solve this problem, I'll tell you how to restart the client safely without restarting the whole system).
In little time we'll know if your problem is due to the distro or not.
OK Dannielle,
Boinc Manager came-up, and wants me to attach to project?Have this output, my Climate chunk is missing in this Manager, so Boinc itself didnot start?, will show you also print out from new manager, well
has same error though, as seen in screen shots.
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ps axu | grep boinc
doug 3048 0.0 0.0 1756 188 pts/0 R 13:46 0:00 grep boinc
bash-3.00$ cat gui_rpc_auth.cfg
93bee0bf9ddd711f2182b4280613f450bash-3.00$ ./run_manager --verbose
send: -1
send: Bad file descriptor
connect: Connection refused
connect: Operation now in progress
ps axu | grep boinc
The Manager was cut off, sorry no output.Looks to me as thier are new files in
"Home", maybe not.Donnot know what happened.
Thanks
Doug
Sorry, will do that again, just looked and saw the wrong commands
Maybe , yes will redo it.I am doing something wrong, I have Rebooted
and done badly 5 times, the last time I Posted was wrong, will give head shake here.
Thanks Danniele
Doug
bash: [doug@localhost: command not found
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ ps axu | grep boinc
doug 3039 0.0 0.0 1756 176 pts/0 R 18:27 0:00 grep boinc
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ps axu | grep boinc
doug 3041 0.0 0.0 1784 484 pts/0 S 18:28 0:00 grep boinc
bash-3.00$ ./run_client -redirectio -dir home/doug/BOINC
chdir: No such file or directory
./run_client -redirectio -dir /home/doug/BOINC
ls al
cat gui_rpc_auth.cfg
./run_manager --verbose
ps axu | grep boinc
I am trying my 8th try, guess either my typing, this last time, using Kedit
to paste the commands in, but having difficuties.Must be my Ditro, sorry I donnot know it better.Will study the command lines that this system uses, its in the MANUAL, again tonight, and I know you need to study also, for your Exams."Good Luck" Dannilele
Sincerely
Doug
bash: 8192: command not found
bash: [doug@localhost: command not found
bash: [doug@localhost: command not found
bash-3.00$ cd /home/doug/BOINC
bash-3.00$ ps axu | grep boinc
doug 4767 0.0 0.0 1788 488 pts/0 S 19:47 0:00 grep boinc
bash-3.00$ ./run_client -redirectio -dir/home/doug/BOINC
Unknown option: -dir/home/doug/BOINC
Usage: ./boinc [options]
-help show options
-version show version info
-exit_when_idle Get/process/report work, then exit
-show_projects show attached projects
-return_results_immediately contact server when have results
-detach_project <URL> detach from a project
-reset_project <URL> reset (clear) a project
-attach_project <URL> <key> attach to a project
-update_prefs <URL> contact a project to update preferences
-run_cpu_benchmarks run the CPU benchmarks
-check_all_logins for idle detection, check remote logins
too -allow_remote_gui_rpc allow remote GUI RPC connections
-gui_rpc_port port for GUI RPCs
-redirectio redirect stdout and stderr to log files
-detach detach from console (Windows)
-dir <path> use given dir as BOINC home
-no_gui_rpc don't allow GUI RPC, don't make socket
-daemon run as daemon (Unix)
bash-3.00$ ls -al
total 8500
drwxr-xr-x 3 doug doug 4096 Oct 28 18:33 .
drwxr-xr-x 38 doug doug 4096 Oct 28 19:30 ..
-rw-r--r-- 1 doug doug 256 May 5 15:36 binstall.sh
-rw-r--r-- 1 doug doug 257 Sep 25 15:09 binstall.sh~
-rwxr-xr-x 1 doug doug 1744688 May 5 15:36 boinc
-rwxr-xr-x 1 doug doug 434320 May 5 15:36 boinc_cmd
-rwxr-xr-x 1 doug doug 6402756 May 5 15:36 boincmgr
-rw-r--r-- 1 doug doug 1264 May 5 15:36 boincmgr.16x16.png
-rw-r--r-- 1 doug doug 1786 May 5 15:36 boincmgr.32x32.png
-rw-r--r-- 1 doug doug 558 May 5 15:36 boincmgr.8x8.png
-rw-r--r-- 1 doug doug 1893 Oct 28 18:33 client_state.xml
-rw-r--r-- 1 doug doug 1893 Oct 28 18:30 client_state_prev.xml
-rw------- 1 doug doug 32 Aug 25 06:20 gui_rpc_auth.cfg
drwxr-xr-x 33 doug doug 4096 May 5 15:36 locale
-rw-r--r-- 1 doug doug 0 Aug 25 06:20 lockfile
-rwxr-xr-x 1 doug doug 61 Oct 28 05:17 run_client
-rwxr-xr-x 1 doug doug 41 Sep 26 11:54 run_client~
-rwxr-xr-x 1 doug doug 44 Oct 11 15:20 run_manager
-rwxr-xr-x 1 doug doug 44 Oct 11 15:20 run_manager~
-rw-r--r-- 1 doug doug 1527 Oct 28 10:27 stderrdae.txt
-rw-r--r-- 1 doug doug 27240 Oct 28 18:33 stdoutdae.txt
bash-3.00$ cat gui_rpc_auth,cfg
cat: gui_rpc_auth,cfg: No such file or directory
bash-3.00$ cat gui_rpc_auth.cfg
93bee0bf9ddd711f2182b4280613f450bash-3.00$ ./run_manager --verbose
send: -1
send: Bad file descriptor
connect: Connection refused
connect: Operation now in progress
ps axu | grep boinc
Got it,
Took a while to see my typing errors, will also put output from Boinc Manager
Here:Sat Oct 28 19:50:00 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sat Oct 28 19:50:00 2006||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
Sat Oct 28 19:50:00 2006||Data directory: /home/doug/BOINC
Sat Oct 28 19:50:00 2006||Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
Sat Oct 28 19:50:00 2006||Memory: 749.28 MB physical, 1.56 GB virtual
Sat Oct 28 19:50:00 2006||Disk: 18.62 GB total, 17.75 GB free
Sat Oct 28 19:50:00 2006||No general preferences found - using BOINC defaults
Sat Oct 28 19:50:00 2006||Local control only allowed
Sat Oct 28 19:50:00 2006||Listening on port 31416
Sat Oct 28 19:50:00 2006||This computer is not attached to any projects
Sat Oct 28 19:50:00 2006||Visit http://boinc.berkeley.edu for instructions
Well, it worked , I think Dannielle?
Got Boinc up and running, connected to local host anyways,just need to "reatach"
to the new accounts?Let me know.I let er go cause "no output" from { ps axu | grep boinc}Looks like my system is not "right" though?I see another error in my typing for auth,cfg should be auth.cfg, Dohh, its late ,will try again.
WTG {way to go}
Doug
Good news also, Climate finnaly "Tricked" some Ram up,LOLWill try command again Sunday morning Danniele.
Thanks
Doug
Sat Oct 28 20:17:43 2006|climateprediction.net|Sending scheduler request to http://climateapps2.oucs.ox.ac.uk/cpdnboinc_cgi/cgi
Sat Oct 28 20:17:43 2006|climateprediction.net|Reason: To send trickle-up message
Sat Oct 28 20:17:43 2006|climateprediction.net|(not requesting new work or reporting completed tasks)
Sat Oct 28 20:17:47 2006|climateprediction.net|Scheduler request succeeded
Sat Oct 28 20:30:31 2006||Rescheduling CPU: project resumed by user
Sat Oct 28 20:30:31 2006|climateprediction.net|Pausing task hadcm3lbm_blbb_25310106_0 (left in memory)
Sat Oct 28 20:30:31 2006|malariacontrol.net beta|Resuming task wu_11_234_10909_0_1466576549_1 using malariacontrol version 543
ID:
1168 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
OK last night I installed big daddy into a virtual machine. It seems to be working ok. Boinc is @ about 9% of D@H wu.
Here is what I did to get it working.
Note that commands are in between "" (do not copy these "" just whatever is inside)
1. Download Boinc using the link at http://docking.utep.edu
2. Once in the download directory... (If using firefox default is /home/<user>/Desktop (substitute <user> by your user name))
3. run the following command:
"chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
4. run the following command:
"./chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
5. run the following command:
"cd BOINC/"
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
7. gui will appear so now just attached to project as usual
Thats it if you follow those steps one by one you will be up and running in 15 minutes. Keep in mind that thats a fresh boinc install. If already installed start at step 6, but I really recommend you to reinstall boinc will assure us that you have those same files. Let me know if you encounter difficulties, try to copy and paste the commands and anything that you have to paste, this way you dont forget any space.
Ok Doug, now there's Memo helping you :)
Anyway, it seems you have done very well yesterday, compliments!
That sequence is correct, as you can see everything is working, if the manager can't connect (I didn't understand) the problem is only the password.
The password is the one in file gui_rpc_auth.cfg
so you only need to do this, if you have problems with the manager claiming for the right password (this is true and useful in many cases, also good for the future)
1) BEFORE starting the client, open the gui_rpc_auth.cfg file in kedit
and delete the old password. Substitute it with an empty string, followed by ENTER (important, every password, the blank one too, is followed by enter).
Save and close the file.
2) start the client and the manager with the sequence
./run_client -redirectio -dir /home/doug/BOINC
cat gui_rpc_auth.cfg (JUST TO SEE IF THE PASSWORD IS THE ONE YOU HAVE SET)
./run_manager --verbose
Should be all :)
Now, and this is always true, if you want to add something like
ulimit etc etc
just stop the client, kedit the run_client script and add the commands in the first lines.
Save and run the client and the manager.
If this works, to create a desktop link to run the client the right way is very simple.
Just to finish the work, anyway now there's Memo and all things will be ok :)
Well, it worked , I think Dannielle?
Got Boinc up and running, connected to local host anyways,just need to "reatach"
to the new accounts?Let me know.
:) I don't understand why the client is not using the old files, did you have to re-attach to projects? For sure there's a way to have the client to behave "the right way", but yesterday I realized I'm having too many problems with my exam... I have to study more than 16 hours per day until november 14th, if not I risk not to pass it...
OK last night I installed big daddy into a virtual machine.
I LOVE brute force approaches :D
Good work Memo, now take Doug to the goal, maybe a clean install is the best thing in this case.
Thank you,
for your "great" teachings, Dannielle, learned plenty about command lines, thanks to your Help/Was nice meeting and communicating with you also, all the best with your great plans.I may do a complete install of system before applying a new Boinc, just need to ask Memo the best "approach".
OK last night I installed big daddy into a virtual machine. It seems to be working ok. Boinc is @ about 9% of D@H wu.
Here is what I did to get it working.
Note that commands are in between "" (do not copy these "" just whatever is inside)
1. Download Boinc using the link at http://docking.utep.edu
2. Once in the download directory... (If using firefox default is /home/<user>/Desktop (substitute <user> by your user name))
3. run the following command:
"chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
4. run the following command:
"./chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
5. run the following command:
"cd BOINC/"
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
7. gui will appear so now just attached to project as usual
Thats it if you follow those steps one by one you will be up and running in 15 minutes. Keep in mind that thats a fresh boinc install. If already installed start at step 6, but I really recommend you to reinstall boinc will assure us that you have those same files. Let me know if you encounter difficulties, try to copy and paste the commands and anything that you have to paste, this way you dont forget any space.
Thanks Memo,
I am thinking the best way to do this is a complete reinstall of this system, so the file system is clean.Will the new download overwrite my old Boinc Files?
If so wonder if Manager problem will continue? I could delete all Boinc realted files in Home , keep home clean.Would take about 3 hours for complete install
setting-up the partitions {lots of fun} now I have done it 6 times.Would deleteing all old Boinc related files be wrong or harmful to do?
Regarding where to download, usually I download too, desktop, cut, and put in home, home is user? What is a virtual machine?Usually I make the download an "executable" and would assume I do the same with this method of install? I am going to do this cause 35
days of no crunching for Docking makes we want it more!
By installing Boinc this way will Boinc start at Boot?, and how do I start it
if not at Boot?And to install the new Boinc, use a Linux Console, or,the Linux
"Shell"?
That is all the info I need to do a good job, for once,{blush}
I am thinking the best way to do this is a complete reinstall of this system, so the file system is clean.Will the new download overwrite my old Boinc Files?
If so wonder if Manager problem will continue?
Only a word about the manager's problem in connecting to the client.
I want to repeat this few concept because this is really a simple problem, due to the manager "not behaving the right way", and it can be solved in few seconds.
You'll have this problem everytime you start the manager by a method different from usual, or reinstall the manager, or reinstall the client, so it's better to repeat it. Moreover you will experience this problem everytime you have to control a remote client.
That file containing the password is created by the client everytime it is started, if there's not one in client's directory. Once the password is created and the manager knows it, you won't need to remember it and ell anything to the manager, unless the file get deleted or you install a new client or manager or start a second instance of client/manager while the first is running. If this happens, the new manager WON'T READ THE FILE and won't remember the password, so you'll have to type it by yourself when the manager asks for it. You can also modify the file and set the password you prefer, but in this case remember that the manager is not smart and will not manage to read the file, so you'll have to tell the manager the new password.
That's all, for now.
ID:
1179 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
OK last night I installed big daddy into a virtual machine. It seems to be working ok. Boinc is @ about 9% of D@H wu.
Here is what I did to get it working.
Note that commands are in between "" (do not copy these "" just whatever is inside)
1. Download Boinc using the link at http://docking.utep.edu
2. Once in the download directory... (If using firefox default is /home/<user>/Desktop (substitute <user> by your user name))
3. run the following command:
"chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
4. run the following command:
"./chmod 777 boinc_5.4.9_i686-pc-linux-gnu.sh"
5. run the following command:
"cd BOINC/"
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
7. gui will appear so now just attached to project as usual
Thats it if you follow those steps one by one you will be up and running in 15 minutes. Keep in mind that thats a fresh boinc install. If already installed start at step 6, but I really recommend you to reinstall boinc will assure us that you have those same files. Let me know if you encounter difficulties, try to copy and paste the commands and anything that you have to paste, this way you dont forget any space.
Thanks Memo,
I am thinking the best way to do this is a complete reinstall of this system, so the file system is clean.Will the new download overwrite my old Boinc Files?
If so wonder if Manager problem will continue? I could delete all Boinc realted files in Home , keep home clean.Would take about 3 hours for complete install
setting-up the partitions {lots of fun} now I have done it 6 times.Would deleteing all old Boinc related files be wrong or harmful to do?
Regarding where to download, usually I download too, desktop, cut, and put in home, home is user? What is a virtual machine?Usually I make the download an "executable" and would assume I do the same with this method of install? I am going to do this cause 35
days of no crunching for Docking makes we want it more!
By installing Boinc this way will Boinc start at Boot?, and how do I start it
if not at Boot?And to install the new Boinc, use a Linux Console, or,the Linux
"Shell"?
That is all the info I need to do a good job, for once,{blush}
Sincerely, and thank you Memo
Doug
If you reinstall boinc do it in a new directory you will need to reattach to all the projects. If you are in the middle of a huge workunit like climate prediction I will advise you to wait till its done.
By installing boinc from scratch you will be able to replicate my results.
I also moved my boinc download to home... don't like to have a mess in my desktop.
To work on start boinc as a service I think you should first get it working this way, I never set up boinc as a service on linux since I run it on machines that are crunching always and they never shut down, but I think daniele has it this way so I think daniele can help us do that when the time comes
As far as virtual machines I can give you this link to start with:
http://en.wikipedia.org/wiki/Vmware
VMWare is the system that I use but there are others... actually you can find references to those at the bottom of that page. If you need to read in another language you can go to wikipedia for that language and I am very positive that it will have the information as well.
ulimit -s unlimited
cd "/home/doug/BOINC" && exec ./boinc $@
[ Read 2 lines ]
^G Get Help ^O WriteOut ^R Read File ^Y Prev Page ^K Cut Text ^C Cur Pos
^X Exit ^J Justify ^W Where Is ^V Next Page ^U UnCut Txt ^T To Spell
Got this far, still have not downloaded new BOINC, today, but later today, am going from line #6, the w/u I tryed after this got further than any other , but, need to complete the commands, tried pushing Y to get a new file, nothing happened, its me, will try and try and try till I get this.Thanks so much for the help from all. especially Memo, Dave, Danniele, and Andre, the w/u got over 5% wheras usual 3.25%, means nada till I do the completed commands.
Thank you
Doug
bash-3.00$ nohup ./run_client
nohup: appending output to `nohup.out'
nohup ./run_manager &
Thinking this may have done something?
Hello,
Can I delete
the "old" Boinc files as to not make a mess? Before I download the new Boinc?
Sincerely
Doug
Dummy Post,
Have to make decision, 30% way through a large Climate "Model", Have never finished one due to trying different Distros.I have 3 run manager files, a Nano
file, with 2 lines, as above, will reboot and do it right, and get a Charm W/U
to run flawlessy.Could put the "new Boinc" in a seperate file, maybe.Plan to keep tying steps 6 through.
Thanks
Doug
Probably debian mantainers choosen such high limits because the debian structure is very coherent and stable, so it's not a problem (in fact it's not).
Sorry if I can't stay more time on the problem, I've a very difficult exam in few days!
No problem. Thanks for the help.
I'm still working on it. I pulled the source of the stable boinc client from CVS and found that when it starts a process, it does a fork and then the forked thread does an execv() to start the new program on Linux. I don't know for sure if the program that is started with execv() gets the limits of the boinc client or gets the default limits of the user it is running as. This could mean that even if the boinc client has unlimited stack, that the application for Docking@Home gets the default limits for the user.
I've tracked down several places in RedHat RHEL3 where some limits are set for processes during startup, based on the user it is running as. I've now modified every one of those I could track down and modified the service startup code in /etc/init.d/boinc. The following files had to be modified.
The ~/.bashrc doesn't have a ulimit command in it since it executes /etc/bashrc.
When it gets the next Docking@Home work unit, we'll see if this fixed the problem. I was able to change the limits for a specific user so that they are in effect at login, so hopefully they'll stick when the Docking@Home application starts, whether it gets the default limits for the user it's running as or inherits them from the boinc client.
Thanks,
-- David
ID:
1187 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Probably debian mantainers choosen such high limits because the debian structure is very coherent and stable, so it's not a problem (in fact it's not).
Sorry if I can't stay more time on the problem, I've a very difficult exam in few days!
No problem. Thanks for the help.
I'm still working on it. I pulled the source of the stable boinc client from CVS and found that when it starts a process, it does a fork and then the forked thread does an execv() to start the new program on Linux. I don't know for sure if the program that is started with execv() gets the limits of the boinc client or gets the default limits of the user it is running as. This could mean that even if the boinc client has unlimited stack, that the application for Docking@Home gets the default limits for the user.
I've tracked down several places in RedHat RHEL3 where some limits are set for processes during startup, based on the user it is running as. I've now modified every one of those I could track down and modified the service startup code in /etc/init.d/boinc. The following files had to be modified.
The ~/.bashrc doesn't have a ulimit command in it since it executes /etc/bashrc.
When it gets the next Docking@Home work unit, we'll see if this fixed the problem. I was able to change the limits for a specific user so that they are in effect at login, so hopefully they'll stick when the Docking@Home application starts, whether it gets the default limits for the user it's running as or inherits them from the boinc client.
Thanks,
-- David
As far as I know a child process contains all of the same info as the parent with the exception of PID and that sort of things. I would assume that execv() will have the unlimited soft stack option, unless I don't remember my OS class. Will check my notes and edit this post to include the things that change from parent to child process in forking...
Thank you, I'm learning a lot by looking at different approaches to the same problem. :) These days I'm studying hard, but sometimes I come here to read.
As far as I know a child process contains all of the same info as the parent with the exception of PID and that sort of things. I would assume that execv() will have the unlimited soft stack option, unless I don't remember my OS class. Will check my notes and edit this post to include the things that change from parent to child process in forking...
I remember the same thing as you, the problem is that in my class nobody spoke about ulimit, too difficult a matter for a simple OS class.
Anyway, it's not really true that those are the only differences between a child and its parent, so it could be that the `unlimited stack` won't be inherited (is it the right word in english?).
Moreover (and anyway :D), I'm pretty sure that ulimit only applies to shell instances and sons, so it could be that a user who has unlimited stack on the login shell (or on the first interactive shell), has unlimited stack on other shell instances too.
Falling back to electronics... I hate numerical filters...
As far as I know a child process contains all of the same info as the parent with the exception of PID and that sort of things. I would assume that execv() will have the unlimited soft stack option, unless I don't remember my OS class. Will check my notes and edit this post to include the things that change from parent to child process in forking...
I remember the same thing as you, the problem is that in my class nobody spoke about ulimit, too difficult a matter for a simple OS class.
Anyway, it's not really true that those are the only differences between a child and its parent, so it could be that the `unlimited stack` won't be inherited (is it the right word in english?).
Moreover (and anyway :D), I'm pretty sure that ulimit only applies to shell instances and sons, so it could be that a user who has unlimited stack on the login shell (or on the first interactive shell), has unlimited stack on other shell instances too.
Falling back to electronics... I hate numerical filters...
The "intellingence" in this thread is wonderful, I donnot understand that much, a little, but it is Paramount that others get the opportunity to read all the
input, ideas, and gained Knowledge.
Thanks Danniele, Memo, David, Andre
Good luck
Sincerely
Doug Worrall
ID:
1197 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
bash-3.00$ nohup ./run_client &
[1] 3054
bash-3.00$ nohup: appending output to `nohup.out'
nohup: cannot run command `./run_client': Permission denied
nohup ./run_client &
nohup: appending output to `nohup.out'
nohup: cannot run command `./run_client': Permission denied
[2] 3055
[1] Exit 126 nohup ./run_client
[2] Exit 126 nohup ./run_client
bash-3.00$ nohup ./run_manager &
nohup: appending output to `nohup.out'
nohup: cannot run command `./run_manager': No such file or directory
[1] 3056
[1] Exit 127 nohup ./run_manager
bash-3.00$
Hello Memo,all
I get this all the time I do-it, seems run_client never "takes", so it is the install.?Guessing, Will study more on this systems commands, due to failures.
Get as far as yes then after first command after getting into Nano.
Each time I have to "reboot" cause it will not attach to network due to wrong password, and donnot understand the way to change that also.Been working this weekend and this week due to a film crew, and,WORK, am unable to try the commands as often as time gives me, but
will not give up.I just saw the " " on run_manager, will do this by hand.
Thanks
Doug Worrall
Have been studying this system, Went to Repository, and installed this xtra
program to compliment the Bash type one, I use.Was thinking that maybe with this,
new commands may be in order.Will try steps 6 trhough now with this "aboard"
Memo.
Documentation for the GNU Bourne Again shell (bash)
Bash is a GNU project sh-compatible shell or command language
interpreter. Bash (Bourne Again shell) incorporates useful features
from the Korn shell (ksh) and the C shell (csh). Most sh scripts
can be run by bash without modification.
Bash offers several improvements over sh, including command line
editing, unlimited size command history, job control, shell
functions and aliases, indexed arrays of unlimited size and
integer arithmetic in any base from two to 64. Bash is ultimately
intended to conform to the IEEE POSIX P1003.2/ISO 9945.2 Shell and
Tools standard.
Bash is the default shell for Mandriva Linux. You should install
bash because of its popularity and power. You'll probably end up
using it.
This package include doc guid examples and manual for zsh.
I have no idea what this may do to help, Just wanted to let you know, I have not, nor, will I give-up."hoping" this will stop those errors right at the beginning of the session.
Hello,
Have a charm w/u running on a a new attached to project Boinc.The charm w/u is at 9.6% done, and 22 minutes in, does this sound correct?Followed all the steps that Memo put out for me, am leaving console up, last time I closed it I lost the Boinc from Running.Checked Ksgaurd, and, Boinc is running and Manager is attached to local host.Have everything saved, but, How do I start this Boinc up again? I fear
turning it off, and losing it?
Hello,
Have a charm w/u running on a a new attached to project Boinc.The charm w/u is at 9.6% done, and 22 minutes in, does this sound correct?
When I had errors in my WUs, no one showed after 11 minutes, most of the times it took 8-9 minutes to stop the execution :)
This one is 14.612% after 40 minutes any take?
Thanks
Dannielle
Doug
You have scored this goal!
GOAAAAAAALLLLLLLLLLLLL
Good job Doug and Memo :D
Thanks Dannielle and Memo,
How would I restart this #2 Boinc, without the Climate Model. which I donnot care about, if this continues to work, and not error out.Should I save the
Konsole session?Still is up now?Have saved all the outputs,etc..,
Its now 17% 50 minutes, are the w/u that stable? and accurate?
How would I restart this #2 Boinc, without the Climate Model. which I donnot care about, if this continues to work, and not error out.Should I save the Konsole session?Still is up now?Have saved all the outputs,etc..,
Its now 17% 50 minutes, are the w/u that stable? and accurate?
I have not followed the last posts because of the exam, so I don't exactly what you have done on your system. Memo will have the right answer to this question.
After solving this "stack's problem" my WUs were all crunched without any errors, so I hope they will behave the same on your system.
How would I restart this #2 Boinc, without the Climate Model. which I donnot care about, if this continues to work, and not error out.Should I save the Konsole session?Still is up now?Have saved all the outputs,etc..,
Its now 17% 50 minutes, are the w/u that stable? and accurate?
I have not followed the last posts because of the exam, so I don't exactly what you have done on your system. Memo will have the right answer to this question.
After solving this "stack's problem" my WUs were all crunched without any errors, so I hope they will behave the same on your system.
Here is the out put from steps 6 on"
ulimit -s unlimited
cd "/home/doug/BOINC" && exec ./boinc $@
[ Read 2 lines ]
bash-3.00$ y
bash: y: command not found
bash-3.00$ nohup ./run_client &
[1] 3034
bash-3.00$ nohup: appending output to `nohup.out'
nohup ./run_manager &
nohup: appending output to `nohup.out'
[2] 3035
bash-3.00$
And now the w/u is 25.7% after 1 hour.Memo, if you are able to read this some time sonn, could you let me know the next steps, I will just have to leave the P.C. crunching these w/u till I get home, She is running quite Warm right now due to the higher temps outside, "For Haloween"Now, 33% for 1 hour and 32 seconds, sound O.K.?
How would I restart this #2 Boinc, without the Climate Model. which I donnot care about, if this continues to work, and not error out.Should I save the Konsole session?Still is up now?Have saved all the outputs,etc..,
Its now 17% 50 minutes, are the w/u that stable? and accurate?
I have not followed the last posts because of the exam, so I don't exactly what you have done on your system. Memo will have the right answer to this question.
After solving this "stack's problem" my WUs were all crunched without any errors, so I hope they will behave the same on your system.
Here is the out put from steps 6 on"
ulimit -s unlimited
cd "/home/doug/BOINC" && exec ./boinc $@
40% now 1 hour 50 minutes, Good
Doug
[ Read 2 lines ]
bash-3.00$ y
bash: y: command not found
bash-3.00$ nohup ./run_client &
[1] 3034
bash-3.00$ nohup: appending output to `nohup.out'
nohup ./run_manager &
nohup: appending output to `nohup.out'
[2] 3035
bash-3.00$
And now the w/u is 25.7% after 1 hour.Memo, if you are able to read this some time sonn, could you let me know the next steps, I will just have to leave the P.C. crunching these w/u till I get home, She is running quite Warm right now due to the higher temps outside, "For Haloween"Now, 33% for 1 hour and 32 seconds, sound O.K.?
Seems you finally got our ulimit fix working :-) That's good news!
Workunits are anywhere in between 2 and 24 hours depending on the machine you are crunching with. A new fast Intel, AMD or Mac might take 2 hours, while a PII might take up to 24 hours. You seem to be somewhere in between with 4.5 hours or so.
Andre
I have to leave again, but, its at 59.162% and 2 hours and 44 minutes, what size are the average Linux w/u?And how long on the average?
Thanks
Doug
____________
D@H the greatest project in the world... a while from now!
Seems you finally got our ulimit fix working :-) That's good news!
Workunits are anywhere in between 2 and 24 hours depending on the machine you are crunching with. A new fast Intel, AMD or Mac might take 2 hours, while a PII might take up to 24 hours. You seem to be somewhere in between with 4.5 hours or so.
Andre
I have to leave again, but, its at 59.162% and 2 hours and 44 minutes, what size are the average Linux w/u?And how long on the average?
Thanks
Doug
Hello Andre,Thank-You
Just got home. 83.91% complete at 4 hours, you look right on que with the time.
How do I quit Boinc and be able to start this one up again.Could transfer my
Manager to desktop, for now, but how would I restart this client and this manager?Ehaaaaa, took 38 days or so of really beating my head against the wall.
Ehaaaaa,
Thank you for all you have acomplished,Also the Console is still "up".Have many options in its menu.Think I need to save it somehow?Last time I quit the console when Boinc Manager was up and running, lost the Boinc Manger and the client running..The console is not using any power at all by running a top on other
console.Would I have to keep doing the commands, or a simplified one?
Ehaaa
Doug
Sincerely
Doug
Seems you finally got our ulimit fix working :-) That's good news!
Workunits are anywhere in between 2 and 24 hours depending on the machine you are crunching with. A new fast Intel, AMD or Mac might take 2 hours, while a PII might take up to 24 hours. You seem to be somewhere in between with 4.5 hours or so.
Andre
I have to leave again, but, its at 59.162% and 2 hours and 44 minutes, what size are the average Linux w/u?And how long on the average?
Thanks
Doug
SUCCESS,
40703 11650 31 Oct 2006 12:45:23 UTC 31 Oct 2006 18:16:45 UTC Over Success Done 16,721.05 16.17 16.17
Ehaaaa, great, did it.
Doug
Thanks Memo, Dannielle,David and Andre
Hello Andre,Thank-You
Just got home. 83.91% complete at 4 hours, you look right on que with the time.
How do I quit Boinc and be able to start this one up again.Could transfer my
Manager to desktop, for now, but how would I restart this client and this manager?Ehaaaaa, took 38 days or so of really beating my head against the wall.
Ehaaaaa,
Thank you for all you have acomplished,Also the Console is still "up".Have many options in its menu.Think I need to save it somehow?Last time I quit the console when Boinc Manager was up and running, lost the Boinc Manger and the client running..The console is not using any power at all by running a top on other
console.Would I have to keep doing the commands, or a simplified one?
Ehaaa
Doug
Sincerely
Doug
Also the Console is still "up".Have many options in its menu.Think I need to save it somehow?Last time I quit the console when Boinc Manager was up and running, lost the Boinc Manger and the client running..The console is not using any power at all by running a top on other
console.Would I have to keep doing the commands, or a simplified one?
The console in that case is a session manager, you can verify this by
ps axu |grep bash
and looking at the 8th column. You'll see a code like this ---> Ss.
If I remember well, the second "s" stands for "session manager".
Generally speaking, when a session manager get terminated, its sons get terminated as well (this behaviour can be changed for a particular process).
You can make a process become a session manager in many ways, the first and more useful in this case is to run it as a daemon (i.e. a service running in background with specific features). This way you can close the console and the client will remain active. Ask Memo and David how to manage services in non-Debian distros. Maybe you can simply try the options of the client -daemon, I don't know if it works. I use the debian subsystem for managing daemons.
I don't understand well what you mean by saving the options of the console, because I didn't read carefully the steps you followed, but if you mean that you want to automatically start the client or something similar, then it's simple and Memo will surely help you. You have to create a little init script, link it to an entry in the kde menu, then use the menu.
Also the Console is still "up".Have many options in its menu.Think I need to save it somehow?Last time I quit the console when Boinc Manager was up and running, lost the Boinc Manger and the client running..The console is not using any power at all by running a top on other
console.Would I have to keep doing the commands, or a simplified one?
The console in that case is a session manager, you can verify this by
ps axu |grep bash
and looking at the 8th column. You'll see a code like this ---> Ss.
If I remember well, the second "s" stands for "session manager".
Generally speaking, when a session manager get terminated, its sons get terminated as well (this behaviour can be changed for a particular process).
You can make a process become a session manager in many ways, the first and more useful in this case is to run it as a daemon (i.e. a service running in background with specific features). This way you can close the console and the client will remain active. Ask Memo and David how to manage services in non-Debian distros. Maybe you can simply try the options of the client -daemon, I don't know if it works. I use the debian subsystem for managing daemons.
I don't understand well what you mean by saving the options of the console, because I didn't read carefully the steps you followed, but if you mean that you want to automatically start the client or something similar, then it's simple and Memo will surely help you. You have to create a little init script, link it to an entry in the kde menu, then use the menu.
Anyway, good job Doug!
Thanks Dannele,
Ran Docking for 24 Hours, Stable w/u , Linux Benchs, are actually used, not used to that,For now will start up Boinc with the console, just noy say yes to save file. have 3 of them, LOL.Tried the console to restart Boinc again, did not say yes to save file, did not work, obviously, Beleive I am learning much latley.
Am finishing the Malaria w/u I had in the Old Boinc, once they are done will
start another new Boinc with Console, run Docking w/u all afternoon and night.
Have to make a decision.Inside by Main Boinc folder in /home/doug/BOINC
are the "Docking files",wheras the old Malria,Climate,Docking,etc.. are outside that File.The new ones inside Boinc file also are run_cliient.save,run_client.save.1, and 4 run_manager files.
I am thinking the best thing to do is a compete install of my system.Then download 1,Boinc,follow Memo,s steps, and whala..
Will be trying a few things, like the new files are in KDE browser also.
Can try clicking on the "approprate" ones, and see if I can get err going that way, doubt it though.Cannot keep making new files,Memo, would you like me to take a screen shot of my BOINC file, and set up thr the URL for it here?Just as above?
Also, I did not get the manager error, when starting up the old way, to clear that cue, assides from the Climate Model.
Sincerely
Doug
I made sure that the client was running with ulimit -s unlimited and it still failed after about 800 seconds. I've now upgraded the boinc client from 5.2.13 to 5.4.9 on the RedHat RHEL3 Machine and will see if that makes a difference tonight.
I noticed that the time when it fails on that machine was about 400 CPU seconds for the old WUs (work units) and about 800 CPU seconds for the new WUs, which you had increased the size on. What does the WU do at the beginning that is size related and what does it do immediately after that? Could this be about the time the C wrapper program starts the Fortran program?
Is the WU trying to change the stack limit? I have noticed that RHEL3 drops the lcap privileges for the new program when you do an SU. I've also noticed that even though the program is running as user boinc, it still has the environment set to say the init file for bash (which is also sh) is /root/.bashrc, which the boinc user does not have access privilege to.
BTW, maybe we should start another thread since this one is getting so long.
Thanks,
-- David
ID:
1240 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Doug,
I was away today and now I been trying to understand what is that you are trying to accomplish. I got lost at the middle, can you explainme again what is that you want, please?
For what I can understand at the moment you are asking that if you close the manager the client will keep running. Not 100% sure but I think the client will keep on running as the client an manager are different processes. Will check tomorrow.
I was away today and now I been trying to understand what is that you are trying to accomplish. I got lost at the middle, can you explainme again what is that you want, please?
For what I can understand at the moment you are asking that if you close the manager the client will keep running. Not 100% sure but I think the client will keep on running as the client an manager are different processes. Will check tomorrow.
Hello Memo,
I tryed getting back to the Boinc that you helped me get going, with no success.Attached to docking
and crunched 24 hours flawlessly, thanks to your method from virtual machine, did not download new Boinc yet, cause, I donnot know how to restart Boinc#2.Lets call Boinc#1 the old Boinc way and Boinc#2, the new way.
I wrote a letter to you at Docking UTEP.I could send a screen shot of the /home/doug/BOINC folder, which has run_client.save, run_client.save.1,4 run_manager files, and the Docking files are in there also.Tried not saying "yes" to save the file, while trying to get it going again, no such luck.
tring to get Boinc#2 running again to crunch Docking again.{want to automatically start the client or something similar, then it's simple and Memo will surely help you. You have to create a little init script, link it to an entry in the kde menu, then use the menu.}Dannielle mentioned this about an init scipt? to restart that instance of Boinc, So I can run Docking?And restart it?
When I clicked off the Boinc Manager, Boinc client stopped also, wheras, when I start Boinc#1, and x out the Manager, it keeps crunching.
Thanks Memo
Good luck Dave
Doug
ID:
1242 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
So new boinc client works good with d@h but you are still crunching the other projects with the old boinc?
If that is the case wait till you finish all work units from other projects and re attach them in the new boinc. Just make sure that in old boinc you set it up so you dont get new work units and thats it.
Did that answer part of your question?
If you start the client the way I told you and close the manager the client will keep crunching. The reason why this is true is because the client is a separate process from the manager.
So new boinc client works good with d@h but you are still crunching the other projects with the old boinc?
If that is the case wait till you finish all work units from other projects and re attach them in the new boinc. Just make sure that in old boinc you set it up so you dont get new work units and thats it.
Did that answer part of your question?
If you start the client the way I told you and close the manager the client will keep crunching. The reason why this is true is because the client is a separate process from the manager.
Sorry,
I am being unclear.I donnot know how to restart new Boinc, I donnot know how to restart it, ?the session, or the new Boinc at all.Sorry, dummy here.I would have to Reatach each time to new project.
Thanks again Memo
Doug
ID:
1244 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
So new boinc client works good with d@h but you are still crunching the other projects with the old boinc?
If that is the case wait till you finish all work units from other projects and re attach them in the new boinc. Just make sure that in old boinc you set it up so you dont get new work units and thats it.
Did that answer part of your question?
If you start the client the way I told you and close the manager the client will keep crunching. The reason why this is true is because the client is a separate process from the manager.
Sorry,
I am being unclear.I donnot know how to restart new Boinc, I donnot know how to restart it, ?the session, or the new Boinc at all.Sorry, dummy here.I would have to Reatach each time to new project.
Thanks again Memo
Doug
and under which boinc you got docking @ home working?
So new boinc client works good with d@h but you are still crunching the other projects with the old boinc?
If that is the case wait till you finish all work units from other projects and re attach them in the new boinc. Just make sure that in old boinc you set it up so you dont get new work units and thats it.
Did that answer part of your question?
If you start the client the way I told you and close the manager the client will keep crunching. The reason why this is true is because the client is a separate process from the manager.
Sorry,
I am being unclear.I donnot know how to restart new Boinc, I donnot know how to restart it, ?the session, or the new Boinc at all.Sorry, dummy here.I would have to Reatach each time to new project.
Thanks again Memo
Doug
I "Had" it working the way you told me, steps 6 on.I just donnot know how to restart that instance of Boinc after a reeboot?
Doug
and under which boinc you got docking @ home working?
So new boinc client works good with d@h but you are still crunching the other projects with the old boinc?
If that is the case wait till you finish all work units from other projects and re attach them in the new boinc. Just make sure that in old boinc you set it up so you dont get new work units and thats it.
Did that answer part of your question?
If you start the client the way I told you and close the manager the client will keep crunching. The reason why this is true is because the client is a separate process from the manager.
Sorry,
I am being unclear.I donnot know how to restart new Boinc, I donnot know how to restart it, ?the session, or the new Boinc at all.Sorry, dummy here.I would have to Reatach each time to new project.
Thanks again Memo
Doug
It worked when I followed your steps
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
7. gui will appear so now just attached to project as usual
And each time I must attach to projects again. if I need a "inut" script..,I dont know what or how "too".
This is what dannielle said:
{I don't understand well what you mean by saving the options of the console, because I didn't read carefully the steps you followed, but if you mean that you want to automatically start the client or something similar, then it's simple and Memo will surely help you. You have to create a little init script, link it to an entry in the kde menu, then use the menu.}
Have cleared my que from old boinc, assides from Climate.
Doug
I "Had" it working the way you told me, steps 6 on.I just donnot know how to restart that instance of Boinc after a reeboot?
Doug
and under which boinc you got docking @ home working?
LOL, was able to start "another" instance of Boinc , your way, will run this w/u, then go into other Boinc and run Climate , back and fourth, at least can crunch for Docking, Hello Sugar.
LOL, was able to start "another" instance of Boinc , your way, will run this w/u, then go into other Boinc and run Climate , back and fourth, at least can crunch for Docking, Hello Sugar.
Sincerely
Doug
Memo,
After I finish this w/u, and I want to start-up "this" instance of Boinc, after ending the session. How might I do that?.By the way, PClinuxOS have a Folding@home team, does not use a Boinc Client, just the way Seti used to be.FYI, thats all.
Well, 2 units done today anyways.I have both boinc client and Manager still operational.Am trying to find way to save, run Climate, Seti, Malaria, Ralph/
for a while. Then get back to docking. Need to find way to not have to connect back to Docking, and rewright extra files.?
Here is what I accomplished this afternoon:
45553 13001 9 Nov 2006 14:39:36 UTC 9 Nov 2006 19:26:51 UTC Over Success Done 16,652.13 17.61 16.20
44205 9490 9 Nov 2006 19:19:00 UTC 9 Nov 2006 22:01:49 UTC Over Success Done 8,566.63 9.06 pending
Well, 2 units done today anyways.I have both boinc client and Manager still operational.Am trying to find way to save, run Climate, Seti, Malaria, Ralph/
for a while. Then get back to docking. Need to find way to not have to connect back to Docking, and rewright extra files.?
Here is what I accomplished this afternoon:
45553 13001 9 Nov 2006 14:39:36 UTC 9 Nov 2006 19:26:51 UTC Over Success Done 16,652.13 17.61 16.20
44205 9490 9 Nov 2006 19:19:00 UTC 9 Nov 2006 22:01:49 UTC Over Success Done 8,566.63 9.06 pending
Doug
Hi Doug, my great friend.
I have little spare time and I can't read the part of the thread I didn't participate to, so would you explain your problem?
Are you running two boinc clients for docking and for the other projects?
Hope all is well with yourself., Yes, I am able to run 2 differenet Boinc,s.One does not have Docking, but has 4 othere experiments, then on the other Bonc
I run Docking, where the ulimit -s unlimited "Works", LOL, Not really a Big
problem.Just need to restart the Docking Boinc by using steps 5 on, here:
5. run the following command:
"cd BOINC/"
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
7. gui will appear so now just attached to project as usual
Each time I start The Docking Boinc "up"
, New files are written, so, just delete them and attach to project "Docking" each time.No Big Deal, will crunch a few docking w/u daily and All is well.
Also run Folding@home, which is a defunct Boinc project, but PClinuxOS has there
own Team, and doesnot use Boinc, nice to see "great" Benchs without BOINC, for some reason also Measured floating point speed 598.05 million ops/sec
Measured integer speed 1626.38 million ops/sec, are my new Boinc Bench,s
, Floating has gotten better by 40 million opps, and integer by 150 opps.
I am sure you did real well for your exams Dannielle, Thank you my friend.
I am sure you did real well for your exams Dannielle, Thank you my friend
:) The exams will be on november 14th, now I'm studying hard.
Regarding your problem, maybe the simplest solution is to set the first client (the one with 4 projects attached) to "don't allow new tasks", and when it completes the current ones you can detach the client from those projects, and reattach the second client. Then you (we) make a shortcut on the desktop and you won't have to use the console. I know that you have climateprediction running on the first client, so it's a problem to complete that WU, but you could start with the other 3 projects. It's not simple to run two clients with the same manager if you are using the same directory with the same user on the same host, I don't even know if the standard client and manager can manage this situation. If they can do it, probably it would be a bit difficult to make the trick work.
WHile I'm waiting for my exams to come to an (happy) end... you are using kde, so you probably have these two programs: Krdc and Krfb. They allow to share a desktop session, so I could work on your system from mine. I've not used them yet, but they could be useful if you have any problems and want me to solve the problem without having to write thousands of lines in this thread. Not a problem for now, I have to study. Bye Doug :)
"Then you (we) make a shortcut on the desktop and you won't have to use the console." This is the answer
Great idea Dannielle,
That is exactly what I need to do, well, "We", LOL.
Thank you so much for your ongoing help.Good luck with the exams, I know you
will do great.Yes, I have "no more work" from all experiments and have only
the Climate Model, and a Seti w/U about 6 hours to finish.
I am also very busy at work , and events I want to volunteer at .So , no biggy
can still crunch for Docking, and finish the Seti, w/u, untill we make the Desktop
shortcut, instead of the Cosole, Wonderful idea Dannielle.
It may have worked prperly this time.Erased all Docking files in home/doug/BOINC,
rebooted, Did steps 5 on, got this message, and no benchmarks, I am sure if I haddnot deleted files, due to attaching each time, I would not have had to reaatach to a project this time.The Console did not ask yes to save, instead it started the Last instance of Boinc Docking.This is the output after starting Boinc Docking up:
Fri Nov 10 23:32:59 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Fri Nov 10 23:32:59 2006||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
Fri Nov 10 23:32:59 2006||Data directory: /home/doug/BOINC
Fri Nov 10 23:32:59 2006|Docking@Home|Project Docking@Home is in state file but no account file found
Fri Nov 10 23:32:59 2006||Application charmm outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||File info outside project in state file
Fri Nov 10 23:32:59 2006||Application version outside project in state file
Fri Nov 10 23:32:59 2006||Workunit outside project in state file
Fri Nov 10 23:32:59 2006||Task 1tng_mod0001_7032_252655_3 outside project in state file
Now I will leave the files as is, and should have success each time.Crunching Docking all night tonight.
:) The exams will be on november 14th, now I'm studying hard.
Good luck on your exams daniele!
Doug, I'm glad things are going better for you!
I'm still fighting with RHEL3, but I might have it beaten. I put an if statement with some specific code for user boinc in the various setup files (/etc/profile, /etc/bashrc, /etc/profile.d/limits.sh) that mess with limits. We'll see tonight. It only gets one work unit per day and processes it immediately.
I'll keep my fingers crossed for everybody :-)
OT: BTW, who won the college football game. I live in Alabama and I think we played UTEP today. I don't usually keep track of football (probably daniele would say American Football). I have kind of gotten hooked into watching the Nextel Cup car racing this year. It's amazing the technology they use in those race cars and also the safety technology. I've seen some amazing crashes and the drivers never seem to get hurt. They hit the wall at 190 mph ( IIRC 100 kph ~= 60 mph), get hit 2 or three other times by cars that couldn't avoid them, and go on to finish the race. A few years ago, just hitting the wall was fatal, but now they often spin around a couple of times until they're pointed in the right direction and continue with the race, possibly without even making a pit stop or going a lap down. I've even seen a driver, who lost the race, drive through the pit area and come flying out and ram the race winner during the victory lap just to show he thought the winner did something wrong. The only injury I heard of this season was to a pit crew member who had changed a tire and didn't get his leg completely out of the way when the jack dropped the car. The safety technology is just unbelievable!!!
Hello Dave,
Trial and error , saved the konsole session last time, then used same Konsole
session to start Boinc. This time will not erase xtra files, and see if it works,
I.M.H.O., it will, ehaaa.Thanks
Regarding UTEP and its football,"American syle" teams name is the Sun Devils.
2½ hours before kickoff
Alpha Lawn, just south of the Whiteman Tennis Center and Parking Structure is the
always enjoyable Tailgate Party. Washington State are playing today against them.
Good luck Sun Devils
have nearly finished the second 4 hour w/u that crunched through the night flawlessly. For more info on the Football team and a calender please copy and
paste this url."Happy Crunching"
http://www.asu.edu/
Doug
ID:
1372 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
Hello Dave,
Trial and error , saved the konsole session last time, then used same Konsole
session to start Boinc. This time will not erase xtra files, and see if it works,
I.M.H.O., it will, ehaaa.Thanks
Regarding UTEP and its football,"American syle" teams name is the Sun Devils.
2½ hours before kickoff
Alpha Lawn, just south of the Whiteman Tennis Center and Parking Structure is the
always enjoyable Tailgate Party. Washington State are playing today against them.
Good luck Sun Devils
have nearly finished the second 4 hour w/u that crunched through the night flawlessly. For more info on the Football team and a calender please copy and
paste this url."Happy Crunching"
http://www.asu.edu/
Doug
Doug what files are you erasing?
As far as I know you don't have to erase anything.
Was a mistake to erase the dockimg files iside home/doug/BOINC, but, by doing this I learned how to "restart" Boinc without reaataching to experiments.What I did was
save the Console session that uses your command lines, now, am able to restart
that instance of Boinc through trial and error, mostly erorrs.I cannot learn unless
I try new ways.
I call up the cosole, open up the saved session, and easily follow step "6"
6.
a) run the following command:
"nano run_client"
b) inside the editor on the top line a type the following:
ulimit -s unlimited
c) hit the keys ctrl x (hold ctrl key and hit x)
d) nano will ask if you want to write the file
press y (y means yes in this case)
e) run the following command:
"nohup ./run_client &"
f) run the following command:
"nohup ./run_manager &"
I donnot have to press "Y" to save that file, seeing as I saved that session.
And it works well, thanks again Memo.
BTW, am I correct that utep football team is the "Sun Devils"?
Thanks Doug
MEMO,
EEEEHHHAAAA, what I have done, is just reload the "Docking" terminal session,
and it works just fine, Thanks Memo, Dannielle, Andre, Dave.
Linux is even more easier than Windows. EHAAA, This thread is now DEFUNc_TO DEFAc_TO, it may help other dummies though. LOL
"Crunching for Docking for 8 hours again", then finish all in Boinc#1.
I have copied the start up script from both instances of Boinc, find the differences to be interesting.
For a Linux Box, and, a start up sript, which one looks the most correct?
Find it quite fascinating, to be able to run 2 types of Boinc, both with
identical Bench,s. without "conflict"?, The system is running Flawlessly.."Cool".
Ciao
Sun Nov 12 16:14:06 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sun Nov 12 16:14:06 2006||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
Sun Nov 12 16:14:06 2006||Data directory: /home/doug/BOINC
Sun Nov 12 16:14:06 2006||Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
Sun Nov 12 16:14:06 2006||Memory: 748.96 MB physical, 1.56 GB virtual
Sun Nov 12 16:14:06 2006||Disk: 18.62 GB total, 17.70 GB free
Sun Nov 12 16:14:06 2006|Docking@Home|URL: http://docking.utep.edu/; Computer ID: 571; location: home; project prefs: default
Sun Nov 12 16:14:06 2006||General prefs: from Docking@Home (last modified 2006-09-30 10:22:27)
Sun Nov 12 16:14:06 2006||General prefs: no separate prefs for home; using your defaults
Sun Nov 12 16:14:06 2006||Local control only allowed
Sun Nov 12 16:14:06 2006||Listening on port 31416
Sun Nov 12 16:14:47 2006|Docking@Home|Sending scheduler request to http://docking.utep.edu/docking_cgi/cgi
Sun Nov 12 16:14:47 2006|Docking@Home|Reason: To fetch work
Sun Nov 12 16:14:47 2006|Docking@Home|Requesting 8640 seconds of new work
Sun Nov 12 16:14:52 2006|Docking@Home|Scheduler request succeeded
Sun Nov 12 16:14:54 2006|Docking@Home|Started download of file 1tng_mod0001_10587_440463.inp
Sun Nov 12 16:14:58 2006|Docking@Home|Finished download of file 1tng_mod0001_10587_440463.inp
Sun Nov 12 16:14:58 2006|Docking@Home|Throughput 435011 bytes/sec
Sun Nov 12 16:14:59 2006||Rescheduling CPU: files downloaded
Sun Nov 12 16:14:59 2006|Docking@Home|Starting task 1tng_mod0001_10587_440463_1 using charmm version 502
Sun Nov 12 16:59:43 2006|Docking@Home|Sending scheduler request to http://docking.utep.edu/docking_cgi/cgi
Sun Nov 12 16:59:43 2006|Docking@Home|Reason: Requested by user
Sun Nov 12 16:59:43 2006|Docking@Home|(not requesting new work or reporting completed tasks)
Sun Nov 12 16:59:48 2006|Docking@Home|Scheduler request succeeded
Sun Nov 12 16:59:48 2006|Docking@Home|General preferences have been updated
Sun Nov 12 16:59:48 2006||General prefs: from Docking@Home (last modified 2006-11-12 15:56:03)
Sun Nov 12 16:59:48 2006||General prefs: no separate prefs for home; using your defaults
Sat Oct 28 05:19:19 2006||Starting BOINC client version 5.4.9 for i686-pc-linux-gnu
Sat Oct 28 05:19:19 2006||libcurl/7.15.3 OpenSSL/0.9.8a zlib/1.2.3
Sat Oct 28 05:19:19 2006||Data directory: /home/doug
Sat Oct 28 05:19:19 2006||Processor: 1 GenuineIntel Intel(R) Celeron(R) CPU 2.40GHz
Sat Oct 28 05:19:19 2006||Memory: 749.28 MB physical, 1.56 GB virtual
Sat Oct 28 05:19:19 2006||Disk: 18.62 GB total, 17.68 GB free
Sat Oct 28 05:19:19 2006|climateprediction.net|URL: http://climateprediction.net/; Computer ID: 473373; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|malariacontrol.net beta|URL: http://www.malariacontrol.net/; Computer ID: 7568; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|Docking@Home|URL: http://docking.utep.edu/; Computer ID: 98; location: home; project prefs: default
Sat Oct 28 05:19:19 2006|boincsimap|URL: http://boinc.bio.wzw.tum.de/boincsimap/; Computer ID: 37849; location: home; project prefs: default
Sat Oct 28 05:19:19 2006||General prefs: from malariacontrol.net beta (last modified 2006-10-25 17:15:35)
Sat Oct 28 05:19:19 2006||General prefs: no separate prefs for home; using your defaults
Sat Oct 28 05:19:19 2006||Local control only allowed
Sat Oct 28 05:19:19 2006||Listening on port 31416
Sat Oct 28 05:19:19 2006|climateprediction.net|Deferring task hadcm3lbm_blbb_25310106_0
Sat Oct 28 05:19:19 2006|malariacontrol.net beta|Resuming task wu_11_34_10204_0_2120349614_0 using malariacontrol version 542
Sat Oct 28 05:19:19 2006||Using earliest-deadline-first scheduling because computer is overcommitted.
Sat Oct 28 05:19:19 2006||Suspending work fetch because computer is overcommitted.
I couldn't follow your successes in the past days, but I'm happy if you managed to solve the problem! Did you manage? As far I can see you are running two clients in two different home directories :)
I couldn't follow your successes in the past days, but I'm happy if you managed to solve the problem! Did you manage? As far I can see you are running two clients in two different home directories :)
TeHe, Yup,
Works great Dannielle.Thanks "ehh".
Doug
ID:
1386 | Rating: 0
| rate:
/
Memo
Forum moderator
Project developer
Project tester
Joined: Sep 13 06
Posts: 88
ID: 14
Credit: 1,666,392
RAC: 0
I am glad you are crunching and that you are learning new things.
I am glad you are crunching and that you are learning new things.
Thanks Memo,
Today, learned that a reboot isnot nessary after ending a session, and also, the session starts, where I left-off, stayed in memory.
Have a concern, have been studying the Linux credits. This is my P.C.s latest
results here:
48527 13852 15 Nov 2006 23:54:47 UTC 16 Nov 2006 5:25:22 UTC Over Success Done 16,303.71 17.94 pending
48481 13840 16 Nov 2006 21:29:33 UTC 17 Nov 2006 2:10:14 UTC Over Success Done 16,330.77 17.97 pending
47492 12800 15 Nov 2006 19:48:35 UTC 16 Nov 2006 0:45:07 UTC Over Success Done 16,286.76 15.88 15.82
46586 13309 13 Nov 2006 1:22:56 UTC 13 Nov 2006 7:41:59 UTC Over Success Done 16,667.08 17.63 15.98
45553 13001 9 Nov 2006 14:39:36 UTC 9 Nov 2006 19:26:51 UTC Over Success Done 16,652.13 17.61 16.20
45329 12651 11 Nov 2006 9:08:56 UTC 11 Nov 2006 14:18:18 UTC Over Success Done 17,022.82 18.00 18.00
44747 12777 12 Nov 2006 21:12:07 UTC 13 Nov 2006 1:54:45 UTC Over Success Done 16,220.52 17.16 16.24
44205 9490 9 Nov 2006 19:19:00 UTC 9 Nov 2006 22:01:49 UTC Over Success Done 8,566.63 9.06 9.06
44136 12628 11 Nov 2006 4:34:03 UTC 11 Nov 2006 9:47:47 UTC Over Success Done 16,479.02 17.43 16.07
43387 11406 1 Nov 2006 1:20:46 UTC 1 Nov 2006 6:45:25 UTC Over Success Done 16,689.32 16.14 14.66
These results worry me seeing I receive less than 5 cobblestones per Hour?
GenuineIntel
Intel(R) Celeron(R) CPU 2.40GHz
Operating System Linux
2.6.17.14.tex1
Memory 748.96 MB
Cache 256 KB
Swap space 1592.34 MB
Total disk space 18.62 GB
Free Disk Space 17.63 GB
Measured floating point speed 614.46 million ops/sec
Measured integer speed 1635.7 million ops/sec
This is the "Best" ever {Floating speed, and, Integer speed} Ran a Benchmark last night when cache was quite full, and very little free Physical memory, which, I just learned, that your system is well configured, {LINUX} when the
RAM is working well."Still a newbie", not completely sure about this comment.
For Linux, are these credits "on par" with other P.C,s that are close to this one? Thanks for any response.