[GE users] Update to 6.1

Andrea Lorenz lorenz at rz.rwth-aachen.de
Mon May 21 08:30:53 BST 2007


    [ The following text is in the "UTF-8" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some characters may be displayed incorrectly. ]

Hello,

> you are trying to upgrade working cluster? This is always dangerous and 
> we recommend to do it not in that way.
Yes, I update a working cluster. Only for long serial jobs were running. 
All other jobs were pending. I have done it serveral times before (e.g. 
  from 6.0u5 to 6.0u6) without any problems.

> How are you performing your updates? The upgrade procedure does not 
> change the databases in case of classic and BDB spooling (non RPC 
> Server). If you're using the BDB Server on Solaris64 the BDB RPC Server 
> binray will be exchanged.
> Only the rc-files and settings files will be recreated.
We use BDB spooling without a BDB Server. The master is running on lx26-x86.
I stopped all the daemons. On the execution hosts were jobs still 
running I call sgeexecd with argument softstop. There I rename the 
SGE_ROOT directory. Then I update the software and restarted the daemons 
again.


>>> Did you delete the qmaster's spool directory??
No, I did not remove any in the spool directory. The queue configuration 
is still available. But the jobslist was empty.

Now all our users have resubmited there jobs.

Andrea



---------------------------------------------------------------------
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