climateprediction.net home page
Restarting - message timeout

Restarting - message timeout

Questions and Answers : Windows : Restarting - message timeout
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile pschoefer

Send message
Joined: 21 Dec 04
Posts: 4
Credit: 480,331
RAC: 0
Message 27506 - Posted: 25 Mar 2007, 20:42:28 UTC

I\'m writing on behalf of a team mate of mine.
Since he uses BAM! and BOINC 5.8.16, he gets messages like in the following extract:
2007-03-24 11:19:26 [climateprediction.net] Restarting task hadcm3ohe_2135_05743059_0 using hadcm3 version 515
2007-03-24 12:57:15 [---] Restarting hadcm3ohe_2135_05743059_0 - message timeout
2007-03-24 12:57:15 [climateprediction.net] Restarting task hadcm3ohe_2135_05743059_0 using hadcm3 version 515
2007-03-24 13:26:46 [---] Restarting hadcm3ohe_2135_05743059_0 - message timeout
2007-03-24 13:26:46 [climateprediction.net] Restarting task hadcm3ohe_2135_05743059_0 using hadcm3 version 515

He says that before he started BAM! some Einstein-Units and the Climate model were running rotatory, but now there\'s only the Climate model running.
Here\'s his host: #517167.
ID: 27506 · Report as offensive     Reply Quote
Profile Keck_Komputers
Avatar

Send message
Joined: 5 Aug 04
Posts: 426
Credit: 2,426,069
RAC: 0
Message 27512 - Posted: 25 Mar 2007, 23:54:32 UTC

It probably has nothing to do with BAM! Most likely the client does not think it has time to finish the CPDN task before the deadline so it has suspended work fetch for the other projects. Updating the client may have caused the change in behavior, there were some major changes to work fetch and CPU scheduling in the 5.8.x clients.

The other possibility is that he did not attach to einstein in BAM!. Sometimes the account managers will detach any projects not attached using the account manager.
BOINC WIKI

BOINCing since 2002/12/8
ID: 27512 · Report as offensive     Reply Quote
Profile pschoefer

Send message
Joined: 21 Dec 04
Posts: 4
Credit: 480,331
RAC: 0
Message 27520 - Posted: 26 Mar 2007, 7:45:10 UTC - in response to Message 27512.  

It probably has nothing to do with BAM! Most likely the client does not think it has time to finish the CPDN task before the deadline so it has suspended work fetch for the other projects. Updating the client may have caused the change in behavior, there were some major changes to work fetch and CPU scheduling in the 5.8.x clients.

Yes, I checked his stdoutdae.txt, and the new behaviour started after upgrading and before BAM!. But he has still one EinsteinWU waiting to run.

The other possibility is that he did not attach to einstein in BAM!. Sometimes the account managers will detach any projects not attached using the account manager.

There\'s a weird thing in his client_state.xml: I cannot find the <attached_via_acct_mgr/>-tag for any project. But BAM! does not detach the projects, so they seem to be registered in BAM!.
ID: 27520 · Report as offensive     Reply Quote
Profile Keck_Komputers
Avatar

Send message
Joined: 5 Aug 04
Posts: 426
Credit: 2,426,069
RAC: 0
Message 27542 - Posted: 27 Mar 2007, 4:30:48 UTC - in response to Message 27520.  


There\'s a weird thing in his client_state.xml: I cannot find the <attached_via_acct_mgr/>-tag for any project. But BAM! does not detach the projects, so they seem to be registered in BAM!.

That is actually the proper behaviour. However several combiniations of server, account manager, and client versions caused projects to detach when not attached via the account manager. Sometimes even a detach and reattach for all projects when starting the account manager happened, but that bug was fixed quickly.
BOINC WIKI

BOINCing since 2002/12/8
ID: 27542 · Report as offensive     Reply Quote

Questions and Answers : Windows : Restarting - message timeout

©2024 climateprediction.net