[GE users] high CPU load for sge_qmaster

Sean Dilda agrajag at dragaera.net
Tue May 10 01:21:12 BST 2005


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

christian reissmann wrote:
> Hi Sean, Stephan
> 
> FYI: Your using 60u3 in CSP mode, the CSP framework was reworked for the
> 60u4 release. May it be an option for you to switch to 60u4?

I plan on switching to 6.0u4 at some point.  However, even if 6.0u4 was 
final today, I wouldn't be able to do the cluster-wide outage to upgrade 
until at least mid-June.
> 
> Shutting down the qmaster in CSP mode for 60u3 requires resyncronizing
> between master and exec daemons. This results in several error messages
> and can take (in the worst case) up to 7 minutes. As you see from the log
> the execds reconnected within one minute ;-)
> 
> This is a protocol inherent problem for 60u3 and below.
> 
> Did all execd reconnect again? Was the master still busy for a longer time
> afterwards?

Yes, the execd's reconnected again.  I'm not sure what you mean by 'busy 
for a longer time'.  The thread is still using 100% cpu usage if that's 
what you're asking.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list