Re: [HTCondor-users] condor_status inconsistency

[John M Knoeller <johnkn@xxxxxxxxxxx>]

As long as this is transient, this is normal. 

The slots don’t all update in the collector in lock-step.

 

From: HTCondor-users [mailto:htcondor-users-bounces@xxxxxxxxxxx] On Behalf Of Jon Bernard
Sent: Sunday, October 30, 2016 5:50 PM
To: HTCondor-Users Mail List <htcondor-users@xxxxxxxxxxx>
Subject: [HTCondor-users] condor_status inconsistency

 

Hi,

 

I've noticed what seems to be an inconsistency in the output of condor_status for dynamically partitioned nodes. If I do

 

condor_status -af Machine Cpus 'sum(childcpus)' state childcpus name childname

 

then for a number of machines I get something like this:

 

node0620 25 3 Unclaimed { 1,1,1 } slot1@node0620 { "slot1_1@node0620","slot1_7@node0620","slot1_8@node0620" }

node0620 1 undefined Claimed undefined slot1_1@node0620 undefined

node0620 1 undefined Claimed undefined slot1_7@node0620 undefined

 

where one of the dynamic slots doesn't appear in the output.

 

The discrepancy is transient, and as far as I can see happens for all the nodes in the pool.

 

This is with 8.5.6.

 

Any ideas?

 

Thanks,

Jon