Current incidents
Currently, it is not possible to run interactive jobs (salloc) from login-1 on Betzy. It appears that the command is hanging, waiting indefinitely on resources:
salloc -A nn9999k -t 10 -N 8 salloc: Pending job allocation 1754891 salloc: job 1754891 queued and waiting for resourceswhereas in fact, the job has already failed. It is just the sacct command hanging.
We are investigating. In the mean time, use login-2 or login-3 for interactive jobs.
There’s currently an issue with filesystem on NIRD SP, which might affect some dependent services. We are working on resolving this.
Apologies for the inconvenience this might cause!
NRIS Team
Information notices
The /cluster/work storage area on Olivia is filling up. Users have been asked to cleanup unused files, but some users might be experiencing issues with jobs getting No space left on device when trying to write data.
We see that some users have already started cleaning up, but it might take some time before this is fully in order. For the time being, please use the $SCRATCH area to write temporary job files (NB: the designated scratch area for a job is deleted when the job ends, so remember...
We have noticed an issue with incoming tickets to [email protected], where lines of text are “disappearing” upon arrival (either partially, or whole email - we receive empty mails). Based on some internal testing and testing from users (thank you!), we have discovered that it occurs when:
- Text is formatted before sending: Specifically text which is between lines of
#####s, and even some markdown (though unpredictable interval) formatting. - Email that is forwarded to us, alongside with new...
Scheduled maintenances
The yearly maintenance stop of Saga will happen in week 46: Monday November 9 at 08:00 – Friday November 13 at 16:00.
During the stop, the firmware, operating system, storage system and queue system will be upgraded.
If possible, we will open up login nodes, with access to files in the project areas, before the maintenance is over.
Jobs in the queue that would not be able to finish before the maintenance will not be scheduled until after the maintenence is over.
Due to ongoing power and cooling work in the machine room where Olivia is located, we have to shut down all compute nodes 2026-11-11 at 11:00. We expect to be back online again no later than 08:00 the following morning.
As last time, we expect to be able to have the login node and the file systems online during the maintenance.
We are sorry for the inconvenience.