Advanced search

Message boards : Number crunching : [coproc] message spam in boinc log

Author Message
Carsten Milkau
Send message
Joined: 7 Feb 09
Posts: 11
Credit: 5,723,515
RAC: 0
Scientific publications
Message 44267 - Posted: 23 Aug 2016 | 13:27:37 UTC

I get lots of messages from GPUGRID in the main boinc log ($BOINCHOME/stdoutdae.txt). These two lines repeat every minute:

23-Aug-2016 15:18:51 [GPUGRID] [coproc] NVIDIA instance 0; 1.000000 pending for e35s5_e17s8p0f513-GERARD_FXCXCL12RX_2353730_1-0-1-RND8142_2
23-Aug-2016 15:18:51 [GPUGRID] [coproc] NVIDIA instance 0: confirming 1.000000 instance for e35s5_e17s8p0f513-GERARD_FXCXCL12RX_2353730_1-0-1-RND8142_2

These are not really telling, but look like some heartbeat request/response to me.

They take up 80% of the log on their own, which is quite noisy. Would it be possible to report these in the WU logs instead of the main boinc log (e.g. by a watchdog/wrapper process)?

Profile Retvari Zoltan
Send message
Joined: 20 Jan 09
Posts: 1960
Credit: 12,622,269,019
RAC: 6,563,983
Scientific publications
Message 44268 - Posted: 23 Aug 2016 | 13:42:55 UTC - in response to Message 44267.

These messages showing up because you've turned on coproc_debug in the cc_config.xml file.

Post to thread

Message boards : Number crunching : [coproc] message spam in boinc log