Dear Discovery users,
In order to facilitate the annual shutdown, we put a reservation on all the cluster nodes to start on Friday, June 7th.
at 00:00:00 on 2019-06-07 all running jobs will be terminated.
No jobs that cannot be completed before that point in time will be scheduled to run.
At this time this only affects jobs submitted to partitions with no time limit. If you run jobs on the PI partitions please set the time to a duration that will not exceed 2019-06-07T00:00:00 - (Current time). If you do not do that
your jobs will not run until the shutdown is over.
If your jobs are pending due to nodes unavailable please update the time limit on them using scontrol update job <JOBID> —time=<duration before 6/7>
As I mentioned earlier the default unlimited time is problematic, for maintenance, and even for normal operations. While we can leave the MaxImum Time Limit unlimited a default should be set which can be exceeded on a per-job basis.
Thank you and best regards!
The RC Team