[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Condor-users] logEvictEvent with unknown reason (108)



Dave,

Yes, your basic interpretation is correct. Also, turning on full debug for the startd will increase log traffic, so if you don't think you can get useful results from doing it on an individual node (in the local condor config file or via condor_config_val), then perhaps we can try to debug the problem by just looking at the machine and job ClassAds (using condor_status -l and condor_q -l as close to the time when the problem happens as possible). If you would like to do that, I recommend sending this information to condor-admin@xxxxxxxxxxxx

--Dan

David A. Kotz wrote:

Is my basic interpretation correct?  Did the negotiator match a job to a
machine, which then rejected the job because it felt it was unable to
meet the requirements of the job?

I'm nervous about going into full debug mode.  This will affect hundreds
of machines.  I just grabbed one of the matches from the log, but it's
happening with lots of them.  I'm worried about how much log traffic
will be generated if I switch all 400+ machines in this pool to full
debug.  They all write to a shared filesystem on a NetApp Filer.  Do you
think there's any real need for concern?

- dave


On Wed, 2005-02-16 at 14:05 -0600, Dan Bradley wrote:


David A. Kotz wrote:



2/16 03:42:59 vm2: Got activate_claim request from shadow
(<128.83.144.225:39738>)
2/16 03:42:59 vm2: Job Requirements check failed!



There's the problem. Now, to find out more about why the job requirements are failing to be met, you need to turn on full debugging in the condor config file:

STARTD_DEBUG = D_COMMAND D_FULLDEBUG

--Dan


_______________________________________________
Condor-users mailing list
Condor-users@xxxxxxxxxxx
https://lists.cs.wisc.edu/mailman/listinfo/condor-users