No results - or no credits
24 Feb 2007 21:31:49 UTC
Topic 192478
(moderation:
My Linux machine has been beavering away as fileserver and BOINC client, unattended, for weeks. I just checked, and it seems that Einstein has given it no credit for the last nine tasks. Only ONE of them even shows in pending credit - work unit 28820714.
I updated recently to BOINC v 5.8.8 - is that a known buggy one, or is Einstein ill?
My Windows machine seems to be doing better - that's got credit for all but three of the last 16 tasks.
Does Einstein have something against Linux, or should I be doing an uninstall/reinstall again?
Keef, Essex or Norfolk, England
No results - or no credits
)
The credits you are missing are probably just "hidden".
If you click "show all hosts" in your host listing, you will see a second P3 Linux machine.
When you upgraded, it has created a new host entry. You could go to your hosts page and choose "merge", maybe it lets you combine the host with its "alter ego".
The new instance of that host has 4,630.70 credits btw, results that have been validated and merged into the science database get purged from the BOINC work database after a while - but the credits stay of course.
Your Linux box shows no active results :
Your computer has 29.48MB of memory, and a job requires 57.22MB
No work sent
(there was work but your computer doesn't have enough memory)
Maybe the setting for "Use no more than xxx% of total virtual memory" is not OK?
Your box has enough installed so you might have to allow some more.
RE: The credits you are
)
That is an earlier machine, no longer in use. There are only two machines connected now - the Windows desktop (octavious) and the Linux fileserver (fedora-vin).
The Linux one has completed nine work-units since the 10th February: only one of them shows in the list, and no credit was given for any of the other eight.
Indeed, but it's been at 4630 since 10th February, despite the work it's done.
Ah! That may explain it! Thank you! I'll go and have a dig in the settings. I suspect something changed when I loaded the latest version of BOINC.
Keef, Essex or Norfolk, England
RE: Your Linux box shows
)
This is odd! Where did you see the bit about 29.48MB of memory? The machine has 528MB of RAM, and the config says use up to 75% of that for BOINC. I turned it up to 90% but Einstein still wouldn't download any work-units to the Linux machine.
At the moment, the machine is doing lots for Rosetta, one for SETI, and none for Einstein or Malariacontrol.
Am I missing something?
Keef, Essex or Norfolk, England
Check the server logs on your
)
Check the server logs on your host page ("last contact" column), that's where I have seen this low memory value and the complaint that it was not enough.
I am not using the latest BOINC version so I cannot tell if there's any local setting that reduces the allowed RAM.
Does the Linux "top" command show something that eats your memory maybe?
hmmm, now it says :
2007-02-26 01:03:26.1580 [PID=9780 ] [CRITICAL] Caught signal 15 [scheduler ran 119.976136 seconds]. Exit(1)ing
I wonder if there's something wrong on server side.
You could try to update and then check your host ID in the "last contact" link, maybe it was a temporary problem as other hosts seem to have received work.
The prefs for that host seem to be the "work" ones btw. (venue setting)
It seems to be fixed, your P3
)
It seems to be fixed, your P3 has pulled a workunit
RE: It seems to be fixed,
)
Thanks.
Yes, but it's not getting any results, and none of the pending workunits that have disappeared off the list have been credited. There's no Einstein credits on the P3 since 10th Feb, although there are 10 completed units waiting. Something's wrong. I'll try resetting it. Otherwise, I think I'll pull it from Einstein.
Here's the statistics screen for Einstein on the P3:
Keef, Essex or Norfolk, England
Are we talking about the same
)
Are we talking about the same P3? I see only one workunit pending :
http://einsteinathome.org/host/770357/tasks
The Tualatin 1133 does not have any workunit at all and has not contacted the project for months.
Einstein did have server trouble which might have caused some weird behaviour, but that is fixed. I don't think that you have to reset the project in order to make it run smooth from now on.
Is it possible that you lost unreported workunits when you updated your BOINC client? I might be wrong but I think the one you are using is quite new.
Well something isn't adding
)
Well something isn't adding up here, but the host in question has completed and reported the last 2 results it got successfully with one granted and one pending, so it would seem there is no need to reset the project.
@Keith: If I'm reading your posts correctly, you're saying you had the 8 results in question showing on your pending list, but now they are gone with no credit granted? If this is the case then the only possible explanation is they were lost due to the backend issues the project was having for some reason, although this would be the first report I've seen of this happening.
Unfortunately, unless you have the WUID's logged somewhere there probably isn't much which can be done at this point, including even investigating what actually happened.
Alinator
RE: Keith: If I'm reading
)
No, I don't bother to write down the WUIDs - I just sit back and let BOINC get on with it. This (P3) machine is my fileserver, and crunches BOINC while doing nothing else (which is most of the time).
The graph above shows the blobs for the units that weren't credited. Not worth worrying about, I guess.
I reset the project anyway, and that broke the spell - I got one WU credited.
Keef, Essex or Norfolk, England
Well, good luck then for all
)
Well, good luck then for all future results :-)