Docking@Home Support: Frequently Asked Questions

General Questions

Charmm Questions


General Questions

Why am I getting an unrecoverable error for result 1tng_mod0001_xxx_xxxxxx_x?

This error has been fixed since charm 5.02. It has been reported on Windows machines only. The error was caused by a random seed that is not properly doing what it is supposed to be doing. It writes to one of the output files over and over; this output file is normally about 2kb in size, was growing much bigger than 1 MB, and that is something the boinc client doesn't like.

I am continuously getting the message "There is work, but it has been committed to other platforms". Why is this happening?

BOINC uses an infeasible flag for the results in the shared memory segment. The problem with the infeasible flag is that if a computer that is too weak comes in and requests work it sets everything to infeasible in the shared memory segment. This in turn means that the new computer that comes in looking for work sees that everything is infeasible and just grabs the first workunit they find. Because D@H is using homogenous redundancy (HR) this means that instead of work that has already been designated to a particular os/processor being sent to that computer - any result will be sent to that computer. This causes more workunits to be quickly assigned to a particular HR category and although we eventually work through the shared memory segment results that are marked infeasible, you still have to deal with the issue of most or all of your workunits in the shared memory segment being marked with a HR. This usually results in one or more types of computers not being able to receive work and the 'work is available for other platforms' message being sent out. Please be patient: at some time there will be work again.

My machine is continuously writing to the hard drive, why is this happening?

We are still working on the excessive writing issue. It might take a couple of days to get our checkpointing fixed and working correctly: currently checkpointing is done without looking at the user preference for disk writes.

After the application stops with an error on my machine, there are huge file uploads (sometimes up to 35 MB or so). Why is this happening?

See this thread for more details. The large file that gets uploaded is called charmm.out and is basically the log file that the application uses to write information to. Normally, when the application exits without error, this file only contains the string \'NO ERRORS\', but in the case of an error, we would like to have this file back on the server so that we can better find out why our application was not doing what it was supposed to do. We realize this file can grow very large and we have enabled this feature only for this test phase where we need as much information as possible to solve the problems we find with our app. Later on, this uploading will be disabled or we will only send back the minimum amount of info. One of the main reasons that charmm.out can grow really large is if your machine crashes or freezes when the application is in the middle of writing checkpoint data to disk. After BOINC restarts the application, it will crash, because it finds incomplete data in the file percentdone.str. This can even occur at 80 or 90% of crunching time. We are looking at how to make checkpointing a more atomic operation so that this will not happen as much anymore.

Why are results in a workunit computed by different app versions (5.03 and 5.04)? Because of this my results are invalid.

According to the BOINC developers this is the normal behavior, but in our case not necessarily the best behavior. We will try to find a solution for this problem soon so that it doesn't happen anymore after future app upgrades. When all the distributed results have returned this problem should not occur anymore.

Can I run Docking@Home on my computer with the 64-bit BOINC core client?

Yes you can. The application you are downloading is an exact copy of the 32-bit version though, which means that you shouldn't expect any speedup because of the fact you are running 64 bits.

How is Docking@Home using Homogeneous Redundancy?

At the moment we have the following configured for homogeneous redundancy (HR):
  • Windows/Intel PII/III
  • Windows/Intel P4 and up
  • Windows/AMD K6
  • Windows/AMD K7 and up
  • Linux/Intel P4 and up
  • Linux/Intel PII/III and AMD K7 and up
  • Linux/AMD K6
  • Mac/Intel
  • Mac/PPC (not used yet)
Some of the Intel Macs seem to give different results (maybe as a consequence of having different Intel architectures?), but we haven't found the exact HR rules needed yet. This means we have to change our HR continuously (and thus the boinc source code) as this is not configurable as yet.

Charmm Questions

What is the current version of charm running on Docking@Home?

All of our current app versions can be found on the Apps Page.

Why are Windows machines running Charmm 5.03 and the others 5.02?

This is due to a bug fix that was affecting only windows machines.

Starting Charmm shows a run error. What is the problem?

This means that there is a communication problem between BOINC (in your computer) and the science application (project's server). Heartbeat is used to verify that both sides are running, most of the time the BOINC client just restarts the WU from the last checkpoint. For more information you can check: http://boinc-wiki.ath.cx/index.php?title=No_heartbeat_from_core_client_-_exiting.

I have been receiving a message on my Linux machine where charm exits with code 1. Is there a fix to this problem?

This has to do with the stacksize setting on your machine which is for some distros (SuSE 9.3 and 10 for example) set to unlimited and for others (FCx, Ubuntu, etc) set to a limited value like 10240. On the Bash and K shells your setting can be seen by typing 'ulimit -s' in a terminal. With TC shell the command is 'limit'. To make the Charmm 'exit 1' errors go away, please set the stacksize to unlimited using the command 'ulimit -s unlimited' before you start the BOINC manager or add it to your run_manager.sh and/or run_client.sh start scripts in the BOINC directory. For Ubuntu, if you install BOINC from their repository, please put this line in the script /etc/init.d/boinc-client. This is not saying that Charmm will use all of your memory (it won't), but it gives us a little bit more space to do our simulations correctly and without errors. Please let us know if this does not work for you. Of course don't forget to restart your BOINC manager and resume the D@H project on your boincmgr in case you suspended it before.

Can't find what you are looking for? Check out Other Resources

Go back to Docking@Home Support