[torqueusers] Checkpoint test fails: bug or misconfiguration?
glen.beane at gmail.com
Thu Mar 19 09:02:39 MDT 2009
I wouldn't use BLCR support in torque 2.3.x
it was basically redesigned for the next version of TORQUE (2.4) and
really shouldn't have been included in the 2.3 release. I think there
had even been talk of removing it from the 2.3 branch because it does
not really work properly
I would try a 2.4.0 snapshot if you really want to play with blcr
2009/3/19 Igor Volovichev <salamanca at bk.ru>:
> Hi, all
> I use Torque-2.3.6 compiled with BLCR 0.8.0 enabled.
> When testing as described in
> test #6 fails. The problem is: whatever checkpoint file I choose (using qalter -W
> checkpoint_name=...), only the last one is used . And it is the filename that is submitted
> to restart_script. However "qstat -f" shows changes correctly. I see the attribute change in file
> /var/spool/torque/server_priv/jobs/xxxxxx.JB at host with server running,
> but there is no change in /var/spool/torque/mom_priv/jobs/xxxxx.JB at host where pbs_mom is running - is it correct behavior?
> Could someone advice some solution of the problem? Thanks.
> torqueusers mailing list
> torqueusers at supercluster.org
More information about the torqueusers