Let me paraphrase the line "If started, the message will be ignored, unless it's known that the job is bad. Then the job will be aborted anyhow".

A future client will even cancel jobs before start when past deadline or when it's anticipated not to reach deadline. All for the sake of grid efficiency. Init 3 just speeds up validation and with Beta that's critical... allows earlier abort of the batch.

Ok thanks Sekerob! If the task has already started to run, it will run until it is finished (if it is not a bad WU).

I understand someone got a beta on Linux, so taking the numbers mentioned in places, 663 and 221 (think 211 was a typo), suggest Mac also got a copy... so which platform is this slagger running on?

Is that 8 hours wallclock (elapsed) or 8 hours CPU time? That can have quite quite a difference, but 0.4% is poor for sure! Task properties would tell if you run a 6.6 or 6.10 client.

Maybe park it for a bit in suspend until someone else checks in. Put LAIM (leave application in memory) on, else you might loose progress, which it should not if not having reached first checkpoint (hope those are the rules in place for all clients). Aborting is not helpful at this time as it would only resubmit a new copy to the next tester. Watch the Result Status page quorum for wingmen reports.

Is it strange that nothing strange happens?

I have got a repair one this morning in my Q6600 under Ubuntu 9.10, with version 6.6.41 (the one which says it is 6.6.40 ).

So far so good: total CPU time estimated (by me) at 40-41 hours. Running together with HCMD2 WUs in the other 3 cores.

I find the checkpointing period too long for my taste. Is it a technical constraint or could there be more than 50 checkpoints if we kindly ask?

For information since I have not seen any report yet:
This WU is BETA_ erlc_ a046_ ps0000.
The WU in error which triggered mne shows error "process exited with code 29 (0x1d, -227)" after 2.92 hours.

Cheers. Jean.