Join PrimeGrid
Returning Participants
Community
Leader Boards
Results
Other
drummers-lowrise
|
Message boards :
Problems and Help :
work not reporting as credit
| Author |
Message |
|---|
|
|
|
My new server, a 64 core Linux machine, has been working hard for 3 days, all cores. It has finished a number of large jobs per day, and a few LLR things too. It looks like i'm only getting credit on boincstatsbam for the LLR work. The value should be 20x higher.
____________
| |
|
Reggie Volunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 10 May 14
Posts: 187
ID: 311759
Credit: 128,768,872
RAC: 124,338
                    
|
My new server, a 64 core Linux machine, has been working hard for 3 days, all cores. It has finished a number of large jobs per day, and a few LLR things too. It looks like i'm only getting credit on boincstatsbam for the LLR work. The value should be 20x higher.
This is fairly hard to debug, but here's my guess. I see you have a handful of abandoned tasks, which includes 20 321-LLR tasks. Abandoned tasks are a bad sign to start with, but they are all from the same time about 4 days ago, so it seems that issue was resolved. It looks like you have 2 valid SGS-LLR and a valid 321-LLR. I'm wondering if BAM is seeing these abandoned tasks as valid. There's a few things you can check that might lead to an answer:
1) Are you connected to PrimeGrid through BAM and seperately? If so, that is the cause of the abandoned tasks and possible the credit issue. It may lead to other issues down the line as well, so you should ensure you are only connected to PrimeGrid in one way.
2) How much credit does BAM say this host has? And does it say credit or RAC? I do know BAM uses it's own RAC formula that could show some differences.
3) I'm not familiar with account managers, but if it provides a link to a workunit, user page, or host page, you should use that to verify it really is connected to your account and that tasks/workunits and host IDs match. | |
|
|
|
|
First, thank you for the thoughtful reply.
Using --get_simple_gui_info, i see two connects to primegrid. One via account manager, the other not. I believe the account manager connection is newer, and it seems to reflect recent setting changes in BAM (resource share at 100%). Also shows 475 jobs succeeded.
The second does NOT use account manager. Jobs succeeded: 435, and two jobs failed.
A separate problem, about 4 days ago, is that the LLR tasks were running on all processors, not paying attention to max thread 1 (as per primegrid preferences). So I aborted them all, restarted, and the genefer 18 tasks started up on as many CPUs as I allowed. (Responds to change.)
I still see some LLR tasks occasionally, which i don't understand. I have them all turned off in primegrid prefs. So I suspect you are correct. How do I clear the second connection?
BAM (https://www.boincstats.com/bam/hosts/) shows 49,380 credit yesterday. Doesn't specify RAC or otherwise. Should be very much hither. 500k perhaps. I don't see an explanation of credit vs RAC credit - can you explain?
boinccmd --acct_mgr info does show a connection, and BAM does indicate daily connections. Also, boinccmd --acct_mgr sync shows no errors, and the BAM timestamp updates properly.
So, I think i need to kill the second non-BAM primegrid link. No idea how to do that. Maybe boinccmd --project https://primegrid.com detach, and then make sure BAM still going?
Thank you so much for your reply. Greatly appreciate.
| |
|
|
|
|
i did detach from primegrid directly, now only attached via BAM. And, it's only running genefer 18 tasks, so things might be better. I'll let you know in a bit.
Thank you, Reggie. | |
|
Post to thread
Message boards :
Problems and Help :
work not reporting as credit |
