BOINC not sending out workunits after mass cancel

Message boards : Questions and problems : BOINC not sending out workunits after mass cancel
Message board moderation

To post messages, you must log in.

AuthorMessage
AxiomOmen

Send message
Joined: 22 May 19
Posts: 4
United States
Message 104587 - Posted: 17 Jun 2021, 21:33:49 UTC

Very recently on MilkyWay@home, I deprecated the latest version of milkyway_nbody (v1.80) and un-deprecated an older version (v1.76) to run some optimizations using the older version. However, since the current workunits from my last set optimizations were incompatible with the older version, and because I didn't want to scare all of our volunteers with computation errors, I cancelled all of the old workunits (roughly over 200,000) before the deprecation using the Cancel Jobs by SQL option in milkyway_ops. I then put up 3 new runs using the older version of milkyway_nbody.

After about a day, I discovered that BOINC had not sent out any of the new workunits. I can see that 999 jobs were created for one of the three optimizations (none for any of the others), but no workunits are being sent out at all.

Does anyone have any insight as to what is happening?
ID: 104587 · Report as offensive
Les Bayliss
Help desk expert

Send message
Joined: 25 Nov 05
Posts: 1654
Australia
Message 104588 - Posted: 17 Jun 2021, 21:58:42 UTC - in response to Message 104587.  

Probably best asked at MilkyWay@home.
ID: 104588 · Report as offensive
Profile Jord
Volunteer tester
Help desk expert
Avatar

Send message
Joined: 29 Aug 05
Posts: 15477
Netherlands
Message 104589 - Posted: 17 Jun 2021, 22:23:37 UTC - in response to Message 104588.  

I think he is Milkyway@Home, as to me it sounds like he's talking about deprecating work, not canceling tasks in his cache.

@AxiomOmen, I think you best ask at the BOINC Projects email list https://groups.google.com/a/ssl.berkeley.edu/g/boinc_projects to get help from fellow project admins, and/or the developers. We're mostly helping people with their client problems these days.
ID: 104589 · Report as offensive
Les Bayliss
Help desk expert

Send message
Joined: 25 Nov 05
Posts: 1654
Australia
Message 104590 - Posted: 18 Jun 2021, 1:27:13 UTC

In that case, my apologies.
ID: 104590 · Report as offensive
AxiomOmen

Send message
Joined: 22 May 19
Posts: 4
United States
Message 104598 - Posted: 18 Jun 2021, 14:50:08 UTC - in response to Message 104590.  

I figured out what the problem was. We had an issue with the signature files changing between versions last update. Since the older version was using invalid signature files, MW@home was not sending any executables. We rereleased the old version as a new one with updated signature files, and now I'm receiving nbody workunits.

Thank you all for your time looking into this.
ID: 104598 · Report as offensive

Message boards : Questions and problems : BOINC not sending out workunits after mass cancel

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.