Too Many Wu's Started = Error

Message boards : BOINC client : Too Many Wu's Started = Error
Message board moderation

To post messages, you must log in.

AuthorMessage
The Gas Giant

Send message
Joined: 30 Aug 05
Posts: 65
Message 31355 - Posted: 5 Mar 2010, 2:24:28 UTC

Recent changes to the work schedular means that if BOINC detects deadline issues it will crunch the wu's in deadline problems until the estimated time left for the wu being crunched means it will then complete in time. It then starts on the next wu in deadline trouble. In general terms this is not a problem except when wu's are variable in length or BOINC isn't working for a period of time (like when a computer lock up occurs over night).

This link shows this scenario.

The only problem with this is that if you have 'leave pre-empted wu's in memory' selected then you can end up with many wu's started, not complete and left in RAM causing RAM to be fully used and paging to disk occuring. Not a major problem until virtual RAM runs out as well.

When virtual RAM also runs out I think I've experienced a 'new bug' where I've seen the message "BOINC password error" and the client detachs from the GUI. This has resulted in many wu's erroring out and since BOINC has stopped crunching causes more deadline issues.


Paul.
ID: 31355 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15480
Netherlands
Message 31371 - Posted: 5 Mar 2010, 22:03:46 UTC - in response to Message 31355.  

David added a fix ([trac]changeset:20805[/trac]), which will go into 6.10.37

- client: revisit the domino-effect preemption problem.

Removed my changes of 19 Jan 2010, which didn't work. Added new mechanism: keep track of whether a job J has ever run in EDF. If so, and if another job of the same project and resource type as J is marked as deadline miss, then mark J as deadline miss, so that it won't get preempted.

ID: 31371 · Report as offensive

Message boards : BOINC client : Too Many Wu's Started = Error

Copyright © 2024 University of California.
Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.