Re: [Condor-users] 6.7.12 gridmanager crash

[Jaime Frey <jfrey@xxxxxxxxxxx>]

On Oct 6, 2005, at 12:55 PM, Rod Walker wrote:

> Hi,
> With 6.7.12 I`m seeing frequent crashes of the gridmanager.
>
> 10/6 09:52:48 [27842] ERROR "BaseJob::DoneWithJob called with  
> unexpected
> state IDLE (1)
> " at line 269 in file basejob.C
>
> This is for gt2 jobs to LCG CE's. Some jobs have finished  
> succesfully and
> I haven`t managed to tie the problem down yet. The symptom is so  
> clear I
> thought I`d ask first.

I haven't seen this before. Can you set GRIDMANAGER_DEBUG =  
D_FULLDEBUG in your config file, provoke the crash, and send me the  
resulting gridmanager log file (off the list)?

> Quill works very nicely, which is the reason for upgrading.
> Is there some way to rollback just the gridmanager for now? When I  
> tried
> replacing the gridmanager and gahpserver binaries for older ones I got
> HoldReason = "GlobusResource is not set in the job ad"

The gridmanager and schedd have to be the same version (in  
particular, they must both be pre- or post-6.7.11).

+----------------------------------+---------------------------------+
|            Jaime Frey            |  Public Split on Whether        |
|        jfrey@xxxxxxxxxxx         |  Bush Is a Divider              |
|  http://www.cs.wisc.edu/~jfrey/  |         -- CNN Scrolling Banner |
+----------------------------------+---------------------------------+