Server Problem

[Ananas]

I have 30-40 tasks that were completed about a week ago by both quorum members and have not yet validated. The server status page is currently showing zero workunits waiting for validation. Something is not right.

Same here.

I think we had an issue like that already some time ago, which turned out to be a due-date issue so that the transitioner didn't give the results to the validator yet even though there had been the required number of results.

Back then, Yoyo gave them a kick and they went through the validator without problems.

[Michael H.W. Weber]

Yoyo is away over the weekend. I am sure he will take a look at this as soon as he is back.

Michael.

[yoyo]

The transitioner will do this automatic in 3 days, the next transition time.
yoyo

[kashi]

Thank you all for your replies.

All of my previously fully completed and waiting for validation workunits have now been validated. My number of pending has dropped from over 200 to 85 which are only those waiting for wingmen. Much better, thanks.

[rilian]

Results ready to send 96,447

but on win32 machine / 1.5GB memory i get

Code: Alles auswählen

4/27/2010 2:14:23 AM	RNA World	Requesting new tasks
4/27/2010 2:14:33 AM	RNA World	Scheduler request completed: got 0 new tasks
4/27/2010 2:14:33 AM	RNA World	Message from server: No work sent
4/27/2010 2:14:33 AM	RNA World	Message from server: No work is available for cmcalibrate
4/27/2010 2:14:33 AM	RNA World	Message from server: (there was work but it was committed to other platforms)

[Norman]

the same here

[yoyo]

Results ready to send 96,447

but on win32 machine / 1.5GB memory i get

Code: Alles auswählen

4/27/2010 2:14:23 AM	RNA World	Requesting new tasks
4/27/2010 2:14:33 AM	RNA World	Scheduler request completed: got 0 new tasks
4/27/2010 2:14:33 AM	RNA World	Message from server: No work sent
4/27/2010 2:14:33 AM	RNA World	Message from server: No work is available for cmcalibrate
4/27/2010 2:14:33 AM	RNA World	Message from server: (there was work but it was committed to other platforms)

Do you have selected cmsearch as application?

[rilian]

Do you have selected cmsearch as application?

yes, it is ON for last 5 months

now,
while Results ready to send 92,239

it says "Project has no jobs available"

this host: http://www.rnaworld.de/rnaworld/results.php?hostid=627

actually, it downloads WUs from time to time, but not every time it needs, so machine sits dry

Code: Alles auswählen

4/27/2010 1:22:56 PM	RNA World	Reporting 9 completed tasks, requesting new tasks
4/27/2010 1:23:07 PM	RNA World	Scheduler request completed: got 0 new tasks
4/27/2010 1:23:07 PM	RNA World	Message from server: (Project has no jobs available)
4/27/2010 1:33:17 PM	RNA World	Sending scheduler request: To fetch work.
4/27/2010 1:33:17 PM	RNA World	Requesting new tasks
4/27/2010 1:33:28 PM	RNA World	Scheduler request completed: got 9 new tasks

[Norman]

10 minutes and more to wait for WU´s for only seconds to crunch is to heavy for my understand
ok, the server ...bottleneck.......

[Roland Schneider]

10 minutes and more to wait for WU´s for only seconds to crunch is to heavy for my understand
ok, the server ...bottleneck.......

That is unfortunate, but there is no easy solution to solve this problem for now, we have to wait for the WU's to get longer.
We want to crunch such short WU's on the server itself in the future, but it is not as easy as it sounds to choose, which WU's should be send by BOINC and which not.
Therefore 10 minutes is the minimum for the server to keep it available.

[mxplm]

My understanding of the whole RNA World project is that we scan every fully sequenced organism for specific patterns in their RNA. Most of these organisms are small, e.g. bacteria => short WUs.

If this right, I would assume that we will get as many short WUs in future sub-projects as we do now. In that case, waiting makes no sense. If there is such a strict correlation between DNA-size and WU-duration, I'd suggest to rethink the idea of combining some small jobs (i.e. DNAs) to generate a single WU.

I could try to think of a possible concept for that if yoyo (or someone else) could make the application-source, the boinc program-source (validator, work-generator,...) and some example WUs accessible for me.

P.S.: This way, after every small job, there would be a 'fake' checkpoint, but that's not the focus here. ;-)

[Michael H.W. Weber]

My understanding of the whole RNA World project is that we scan every fully sequenced organism for specific patterns in their RNA.

More specifically: We scan every fully sequenced genome's DNA for its potential to encode ncRNAs.

Most of these organisms are small, e.g. bacteria => short WUs.

The bacterial genomes are not resulting in small WUs, rather it is the plasmids and phages which we also screen. Genome sizes are roughly in the order plasmids < organelles < phages < archaea < bacteria < eukaryota

If this right, I would assume that we will get as many short WUs in future sub-projects as we do now.

Correct.

If there is such a strict correlation between DNA-size and WU-duration, I'd suggest to rethink the idea of combining some small jobs (i.e. DNAs) to generate a single WU.

For validation reasons that is not an option. Also, the outcome of this suggestion would be identical to simply increasing the maximum of allowed number of WUs per CPU core.

We will change our current concept by processing the small WUs on the server - at least in large parts.

Michael.