Questions and Answers :
Unix/Linux :
Multi-processing of a task is not working on Gentoo LInux
Message board moderation
Author | Message |
---|---|
Send message Joined: 6 Nov 05 Posts: 9 Credit: 4,953,270 RAC: 0 |
If I set, say, max jobs = 3 and max CPUs = 4, I get three LHC tasks running with four CPUs each. But, even though time-used goes up and time-to-complete goes down, nothing is actually happening, as shown by /usr/bin/top and by the silence of the system's cooling fans. Do I need to do something else to make this work? Rgds Peter. |
Send message Joined: 15 Jun 08 Posts: 2556 Credit: 255,984,613 RAC: 83,654 |
Peter Your CMS tasks are fine. Regarding ATLAS/Theory: Did you ever read what's printed to your logfiles? https://lhcathome.cern.ch/lhcathome/result.php?resultid=324523265 https://lhcathome.cern.ch/lhcathome/result.php?resultid=324252470 Just a hint: "... CVMFS is required to run ATLAS native tasks..." |
Send message Joined: 6 Nov 05 Posts: 9 Credit: 4,953,270 RAC: 0 |
So what's changed then? I've never had to install net-fs/cvmfs before. (This is a Gentoo box.) I haven't had to go looking around your logs before, either, and I don't see such errors here. Thanks for the fix. LHC jobs are not running at the moment, but I'll come back when they start, if the fix hasn't worked. Rgds Peter. |
Send message Joined: 15 Jun 08 Posts: 2556 Credit: 255,984,613 RAC: 83,654 |
So what's changed then? I've never had to install net-fs/cvmfs before. (This is a Gentoo box.) Nothing has changed. You may have run some vbox tasks before. They run inside a preconfigured virtual machine which has a well configured CVMFS client. The project did that for you. To run native apps a locally installed and well configured CVMFS client is a must. This must be done by the volunteer. See: https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5594 https://lhcathome.cern.ch/lhcathome/forum_thread.php?id=5595 CMS is only available as vbox app, hence they succeed. Theory/ATLAS are available as vbox app as well as (on Linux) as native app. The latter is what your computer ran during the past days and all of them fail due to the same reason - a missing local CVMFS client. The logs I linked to in my previous post are your's, not mine. Hence, this tasklists show your tasks/errors. Theory: https://lhcathome.cern.ch/lhcathome/results.php?hostid=10584191&offset=0&show_names=0&state=0&appid=13 ATLAS: https://lhcathome.cern.ch/lhcathome/results.php?hostid=10584191&offset=0&show_names=0&state=0&appid=14 |
Send message Joined: 6 Nov 05 Posts: 9 Credit: 4,953,270 RAC: 0 |
Nothing has changed. That is clearly untrue. I used to run LHC@home without jumping through any hoops, nor having to install any supporting programs. Now I do have to. You may have run some vbox tasks before. They run inside a preconfigured virtual machine which has a well configured CVMFS client. The project did that for you. Nice of it. To run native apps a locally installed and well configured CVMFS client is a must. This must be done by the volunteer. It would be nice to be told that up front. CMS is only available as vbox app, hence they succeed. Theory/ATLAS are available as vbox app as well as (on Linux) as native app. The latter is what your computer ran during the past days and all of them fail due to the same reason - a missing local CVMFS client. Again, why was I not told that? The logs I linked to in my previous post are your's, not mine. No, they're logs written by your software and held on your machines. My logs are held locally and do not include any of that useful material. Why not? Hence, this tasklists show your tasks/errors. Theory: https://lhcathome.cern.ch/lhcathome/results.php?hostid=10584191&offset=0&show_names=0&state=0&appid=13 ATLAS: https://lhcathome.cern.ch/lhcathome/results.php?hostid=10584191&offset=0&show_names=0&state=0&appid=14[/quote] Rgds Peter. |
Send message Joined: 2 May 07 Posts: 2245 Credit: 174,076,610 RAC: 9,988 |
Please stop the using of your Computer for LHCatHome so long, until you found a solution for this Errors. Your last CMS-Task, finished correct is from 21/08/28. You have more than 500 Tasks errored out. So, a wingman have to finish this tasks. Something must be changed from your side on your Computer. |
Send message Joined: 15 Jun 08 Posts: 2556 Credit: 255,984,613 RAC: 83,654 |
Best you can do Open the preferences page and edit your preferences *): https://lhcathome.cern.ch/lhcathome/prefs.php?subset=project Uncheck "Run test applications?" or keep it unchecked. "Run only the selected applications" Check only SixTrack, uncheck all others or keep them unchecked. Uncheck "If no work for selected applications is available, accept work from other applications?" or keep it unchecked. Uncheck "Run native if available?" or keep it unchecked. Max # jobs: No limit Max # CPUs: 1 Click on "Update preferences" To avoid "...why was I not told that?" Never ask why you don't get any tasks. Instead, check the server status page yourself: https://lhcathome.cern.ch/lhcathome/server_status.php As long as SixTrack shows "0" unsent tasks it really has "0" unsent tasks. Never ask "when will there be new tasks?" since nobody here knows when. Could be tomorrow. Could be in a few weeks. Could be in a few months. *): I know, they are not your's since they are on the project server, but for now let's assume it could be helpful. |
Send message Joined: 6 Nov 05 Posts: 9 Credit: 4,953,270 RAC: 0 |
Peter Is there an easy way to reach these log files? It looks as though I have to remember where they are and append the task ID. No doubt I'm missing something again. Rgds Peter. |
Send message Joined: 15 Jun 08 Posts: 2556 Credit: 255,984,613 RAC: 83,654 |
Is there an easy way to reach these log files? Option 1: To check results (AKA tasks) already reported back to the project On this website navigate to "Project --> Account" from the menu. In section "Computing" either follow the link to - "Tasks: View" or - "Computers on this account: View", then "Tasks" from the computer you want to check. Option 2: To check results currently processing but not yet finished On your computer you find a "stderr.txt" in the ".../slots/n/" directory of each running task. "n" represents the slot number. |
Send message Joined: 6 Nov 05 Posts: 9 Credit: 4,953,270 RAC: 0 |
Thank you. I'll look into those. Rgds Peter. |
©2025 CERN