Message boards : Number crunching : How to automatically end long running tasks?
Message board moderation
Author | Message |
---|---|
Send message Joined: 3 Jul 23 Posts: 1 Credit: 2,856,490 RAC: 6,893 |
Hi, is it possible to use an app_config entry to automatically kill long-running tasks that will just error anyway? I had a task (https://boinc.multi-pool.info/latinsquares/workunit.php?wuid=462370749) run on CPU overnight for nearly 8.5 hours, instead of the usual 20 minutes, only to then be marked as an error. There have been others and they are admittedly quite rare, but it's very wasteful of resources to run for so long to no purpose. Accepting that some tasks may just run for longer than 'normal' but it seems sensible to 'kill' any tasks that take more than say 1.5 or 2 times longer. Thoughts and suggestions much appreciated. |
Send message Joined: 28 Oct 17 Posts: 220 Credit: 59,113 RAC: 14 |
We had investigated into the past this kind of tasks, but we were unable to find a rule to being able to excluded them. Unfortunately Boinc has no too much option for a client, like suspend a long running task. Maybe it could help in setting "change activity" to 25 min", so a long running task will be suspended many times while switching with others. This not resolve the problems, but maybe when you detect it, it has run for less global time and can be killed before running for too much time. |
Send message Joined: 7 Nov 17 Posts: 29 Credit: 13,770,308 RAC: 187 |
This year a lot of new tasks really don't work well. They will never be counted. There are thousands of them. https://boinc.multi-pool.info/latinsquares/workunit.php?wuid=461268905 https://boinc.multi-pool.info/latinsquares/result.php?resultid=530230511 https://boinc.multi-pool.info/latinsquares/workunit.php?wuid=460353174 https://boinc.multi-pool.info/latinsquares/result.php?resultid=531574330 https://boinc.multi-pool.info/latinsquares/result.php?resultid=532228783 Look at the execution time, it is very long. 197 (0x000000C5) EXIT_TIME_LIMIT_EXCEEDED |
Send message Joined: 29 Nov 17 Posts: 15 Credit: 3,120,260 RAC: 8,711 |
This one I aborted after 4hr50min https://boinc.multi-pool.info/latinsquares/result.php?resultid=537617157 One would have to use boinccmd and some scripting to check running tasks. Maybe make a list in the script and compare from an earlier list? |
Send message Joined: 7 Nov 17 Posts: 29 Credit: 13,770,308 RAC: 187 |
Yes. Too many mistakes. Statistics on completed tasks have completely disappeared. I had one host count for 12 hours, then gave an error. In the open part of the project I found several hundred incorrect ones. Until the task generator is fixed, there will be a problem... |
©2024 ©2024 Progger & Stefano Tognon (ice00) & Reese