I can´t get workunits from e@home Servers
20 Dec 2020 1:18:38 UTC
Topic 224258
(moderation:
Yes friends crunchers, i have finished all of my workunits and i can´t get more the servers of einstein@home, what´s happening? Do you have some information, i have seen in the server status, that the worksgenerators are desactivated, can be that?
Grettings in advance.
I see two obvious
)
I see two obvious problems.
1. While your machine has enjoyed success in processing, returning, and getting validation on all 8 CPU tasks, it has generated almost immediate failures on all 35 GPU tasks returned on December 19.
2. You must have a wildly inappropriately high work fetch (cache setting) as the relevant lines in the most recent request log as of just before I am writing read:
2020-12-20 01:20:30.4375 [PID=438 ] [send] CPU: req 2419200.00 sec, 8.00 instances; est delay 0.00
2020-12-20 01:20:30.4375 [PID=438 ] [send] ATI: req 626400.00 sec, 1.00 instances; est delay 0.00
As to why you are not getting work, there would be more explanation in that log had you not hit the "try again" button so soon after the previous request that this log just contains the note:
"Not sending work - last request too recent: 17 sec"
I suspect that had you not hit the button again so fast, the most recent request would log that your daily quota of tasks had been exceeded.
Those 35 errors would have progressively ground down your allowed daily quota to much less than the raw specs on your machine would entitle it to.
If you turn off GPU processing, you won't get more of those errors. If you wait a few hours your eligibility for CPU tasks will resume. As soon as you return any successfully your taily task quote will go back up--quickly.
Then the problem is to figure out why your system can't do these GPU tasks.
Possibly some one else can help on that.
Clues:
Your GPU is reported as: AMD AMD Radeon HD 7700 Series (2048MB)
The Stderr is reported as empty!?!?! (I'm not used to seeing that, ever)
Other information:
Outcome:Computation error
Client state:Compute error
Exit status:68 (0x00000044) Unknown error code
JONATAN: There are enough
)
JONATAN:
There are enough jobs to be downloaded - so don't worry about workgenerators not running.
Do you have the newest driver for you GPU installed?
Have a nice Sunday!
1/6/2021 3:38:00 AM |
)
Since another one of my systems is still getting gpu work. And since this system had a large raft of computation errors which I have fixed the incorrect gpu settings on. I am wondering how long will this system continue to get no gpu tasks messages?
I would rather not start requesting GW gpu tasks.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
Tom M wrote:... I am
)
It will probably take as long as it takes for someone to notice and work out why tasks for the next data file haven't automatically kicked in.
For the GRP search, the data files have the format "LATeah2065L68am.dat" (the latest one in play) and tasks for it finished overnight UTC. The tasks for each file have been lasting about a day and things usually kick over to the next file in the series automatically when the previous tasks have all been issued.
This has happened before and usually by about this time of day in Hannover, the problem is discovered and a new file comes into play. Should happen any time now :-).
Cheers,
Gary.
Gary Roberts wrote: Tom M
)
I did break down and toggle the "run non-preferred tasks when preferred are not available" so I am getting GW gpu tasks now.
Hopefully, that won't interfere with getting GR again when it becomes available.
Tom M
A Proud member of the O.F.A. (Old Farts Association). Be well, do good work, and keep in touch.® (Garrison Keillor) I want some more patience. RIGHT NOW!
This has been a long outage
)
This has been a long outage of Gamma-Ray Pulsar GPU work unit availability. As it is now well past mid-day of a mid-week day in Germany, perhaps the "it will get noticed and fixed" real soon confidence is misplaced.
archae86 wrote: This has
)
Today is Epiphany which is a holiday in Germany and many other places as well. So probably we'll have to wait for tomorrow.
maybe they paused GR work to
)
maybe they paused GR work to fix the upload issues?
_________________________________________________________________________
Harri Liljeroos wrote:Today
)
Thanks for mentioning that - I guess that's the reason.
Einstein4 is not 'down' and there is no maintenance going on. If there were, all the daemons running on that server would show as 'disabled' but they're all still 'running'. If maintenance is planned, there's usually a news announcement so this seems an 'unscheduled' event and nobody has noticed - which seems rather strange.
I've still got work for hosts doing GRP but if they run out completely it would affect stability during the 'feeding frenzy' when work becomes available again. Once the remaining work gets low, I will run a script to suspend all hosts before they actually run out and I wont 'un-suspend' them until the feeding frenzy abates. That way, I wont be adding to the frenzy and there shouldn't be any manual intervention needed by me to 'fix' hosts that get in a bind. I have some experience of these sorts of issues when restarting after a storm induced power outage for example :-).
For the moment, my hosts doing GW will continue as is since there's no problem there. It's 6:15 am here (UTC+10) so I'll be monitoring things for the rest of my day. You never know - someone over there might notice and get things going again :-).
Cheers,
Gary.
According to the server
)
According to the server status page the FGRP5 are showing as 3.3 days left, so maybe they've run out of work for it.
MarksRpiCluster