Server issues?

[nullcoding]

Hi, I'll reuse this thread because it seems silly to start a new one, as I'm sure this is an issue that will be resolved quickly.

I have a 2.53GHz Core i5, so it's got 4 threaded cores and therefore can crunch 4 BOINC tasks at once. Up til recently it was doing this with eOn (and my old Core 2 Duo and Pentium 4 were each doing 2 at a time, as they should). However, after the recent server update (how'd that go, anyway, I didn't see an update but things ARE back online), I get this:

Code: Select all

Thu Oct 21 23:42:42 2010	eon2	Sending scheduler request: To fetch work.
Thu Oct 21 23:42:42 2010	eon2	Requesting new tasks for CPU
Thu Oct 21 23:42:45 2010	eon2	Scheduler request completed: got 0 new tasks
Thu Oct 21 23:42:45 2010	eon2	Message from server: No work sent
Thu Oct 21 23:42:45 2010	eon2	Message from server: This computer has reached a limit on tasks in progress

This comes when I have one task running. Each task takes at most 2.5 minutes and I never get errors in computation, but I'm used to running 4 at once and somehow I can't now.

I'm sure this is just a snag in one of the feeder servers or something, but just letting you know in case it got overlooked...or I'm the only one for whom this is happening.

[chill]

This comes when I have one task running. Each task takes at most 2.5 minutes and I never get errors in computation, but I'm used to running 4 at once and somehow I can't now.

I'm sure this is just a snag in one of the feeder servers or something, but just letting you know in case it got overlooked...or I'm the only one for whom this is happening.

Everything went well in the upgrade, but I haven't restored all of our simulation data from the backup I made. This means that we only have one small simulation running right now. We will soon be back to full capacity.

[Mankka*]

When you are running normally, it would be nice if you could change the limitation of work to at least
2 wu's/core. With only 1/core the computers run out of work before they are allowed to ask for new wu's

Thank you :)

Mankka*

[nullcoding]

I see - thanks chill, good to know.

I understand there's a lot involved in running this project, but we're all pretty eager to crunch for ya - looking forward to getting more tasks as you get more simulations up and running again!

[chill]

When you are running normally, it would be nice if you could change the limitation of work to at least
2 wu's/core. With only 1/core the computers run out of work before they are allowed to ask for new wu's

Much of our work is time sensitive and we want all of our workunits that we make to be worked on and not sit in a client queue. Ideally we would set the limit to 2 wu/core. Right now I am having a problem that after the upgrade boinc will no longer send 1 wu/core but only 1 wu to each client. I am looking into how to fix this. It seems to be related to running the latest version of the server side BOINC software.

[Mankka*]

I would be nice if you could manage to do it :)

Just as information, I'm getting 4 wu's on the compurers with 8 cores...

So you are going in the right direction !

Edit: I'm actually getting 4 wu's for all of my computers (1-8 cores) So the amount of wu's being sent out is still related to the number of hosts, not the number of cores...

[chill]

I would be nice if you could manage to do it :)

Just as information, I'm getting 4 wu's on the compurers with 8 cores...

So you are going in the right direction !

Edit: I'm actually getting 4 wu's for all of my computers (1-8 cores) So the amount of wu's being sent out is still related to the number of hosts, not the number of cores...

For some reason the max_wus_in_progress option on the server side config is no longer specifying that clients may queue up number_of_cores*max_wus_in_progress number of workunits. It is only letting clients queue up max_wus_in_progress number of workunits. I will have to take some time to look into this further.

[Neil Polson]

Just noticed I'm no longer running the 64-bit Linux app. This has happened at a few projects after a server upgrade to the latest version of the server side BOINC software.
Also seem to be able to download as much as we want today?

[graeme]

The number of work units that can be downloaded has been fixed. This could also explain your variation in the credit/hour. Over the next few weeks things should be back to normal, and I expect the credit production rate to stabilize.

[BaDu1967]

Another server issue:

I can't change a eon boinc password via the site since the server update.
After clicking "change password" on the account page, the server responds with:

Not Found
The requested URL /eon2/SECURE_URL_BASEedit_passwd_form.php was not found on this server.

Seems to me the part in the URL that says SECURE_URL_BASE should not be there.
Anyway to fix this soon as im waiting to start crunshing on Eon2?

Senks
BaDu1967

[graeme]

The change password form should be fixed now. Please post back if you see any other problems with the website.
Thanks!

[BaDu1967]

Great!! I can change my password now. This sure is a fast response!

[bhambrice]

I have not received any workunits in several days. I did have one that was stuck in uploading. As was suggested earlier in this thread, I reset the project, then updated. Is the problem on my end, or are there no workunits?
When I do a manual update, the messages say that the scheduler list is being downloaded. Then nothing. It doesn't appear that the update is completing.

Any ideas??

[chill]

I have not received any workunits in several days. I did have one that was stuck in uploading. As was suggested earlier in this thread, I reset the project, then updated. Is the problem on my end, or are there no workunits?
When I do a manual update, the messages say that the scheduler list is being downloaded. Then nothing. It doesn't appear that the update is completing.

Any ideas??

I'm not sure what the problem is. I cannot reproduce it on my computer. If you figure it out you should post your solution here on the forums in case it happens to anyone else.