Dear HPC users,
I have to apologize for the repeated changes regarding the downtime, but we have again received new information from the external engineering contractor with an updated workplan. The downtime is now scheduled two weeks later and it will begin on February 14th at 6:00am. Unfortunately, the required time to complete the necessary work on the cooling system has been estimated with 10 days. Therefore, we expect the cluster to be up and running again on Thursday, Feb 24th.
During the downtime, the login nodes and the storage systems ($HOME, $WORK, …) will be accessible. As mentioned before, you will notice that your jobs may be waiting in the queue with the status ReqNodeNotAvail if the requested runtime is too long with respect to the downtime. If possible you can reduce the runtime of your job with the command
$ scontrol update job <job-id> TimeLimit=<new-runtime-limit>
where you have to replace the <placeholders> with meaningful values. Otherwise, you can leave the job waiting in the queue, it will start automatically once the downtime is over.
Sorry for the inconveniences this may cause.
Best wishes
Stefan