GPFS hang on node “en”
Node en had a GPFS hang. GPFS restartet. Downtime approx. 30 min.
Author Archives: lsz075
Hang of node “en”
Node "en" of the regattas has a hang. We are investigating.
Update 09:45: Node rebooted. No jobs lost.
Update 09:45: Node rebooted. No jobs lost.
/home/fimm crashed
The filesystem /home/fimm crashed for unknown reason on fimm. We are currently investigating.
Users will not be able to log in until this is fixed.
Update 16.40: The filesystem is now up and working again.
Users will not be able to log in until this is fixed.
Update 16.40: The filesystem is now up and working again.
Hang on tre
The tre node has crashed at 15:30, possibly due to disk-issues. We are investigating.
Update 16:30: Machine is now up. Jobs on the "tre" node was lost.
Update 16:30: Machine is now up. Jobs on the "tre" node was lost.
Filesystem hang on tre and en
The GPFS filesystem on tre and en stopped working sometime after Friday 16:30. It was restored (restarted GPFS) again on Saturday 12:00. All jobs lost.
/migrate (but not Bjerknes bcmhsm) is being moved
Due in part to the recent /work failure on tre we are moving the /migrate filesystem. It is therefore unavailable due to being moved from "to" to another machine (bregne). Do not try to use /migrate on "to" anymore.
The moved migrate will be mounted on tre as /migrate and on fimm as /migrate.
We will update this notice with further progress.
Update Thursday 16:50: The move is not entirely complete, but I have mounted /migrate again on tre and fimm. Expect the move to complete in a couple of hours.
The moved migrate will be mounted on tre as /migrate and on fimm as /migrate.
We will update this notice with further progress.
Update Thursday 16:50: The move is not entirely complete, but I have mounted /migrate again on tre and fimm. Expect the move to complete in a couple of hours.
/work filesystem failure on “tre”
The /work filesystem on tre is down (from 07:30), most probably due to disk failure.
We are looking into the issue.
Update (10:55): The /work filesystem is un-recoverable. We are creating a new /work filesystem from the remaining functioning disks. All data on /work is then of course lost.
Update (12:30): All nodes and filesystems are up again.
NB: You must check that you have the necessary folders in /work ready before you submit your jobs.
We are looking into the issue.
Update (10:55): The /work filesystem is un-recoverable. We are creating a new /work filesystem from the remaining functioning disks. All data on /work is then of course lost.
Update (12:30): All nodes and filesystems are up again.
NB: You must check that you have the necessary folders in /work ready before you submit your jobs.
GPFS filesystem problem on fimm
The GPFS filesystem went into a hang at 01:30, related to a power outage, and at 08:00 was forced unmounted/mounted. All running jobs were killed. The filesystem is now OK again.
Matlab upgraded on fimm
Matlab on fimm is upgraded from R14-sp3 to R2007a.
/local/bin/matlab should be used and is now pointing to the new release.
Questions and problems should be sent to support-uib@notur.no
/local/bin/matlab should be used and is now pointing to the new release.
Questions and problems should be sent to support-uib@notur.no
Maintenance stop of tre (to fix cooling)
We need to replace a fan-motor in the cooling unit that cools "tre".
During this maintenence it is expected that the temperature will rise so much that we will be forced to shutdown the regatta nodes.
The maintenance is expected to take place from 08:00 to 10:00 tomorrow, Friday 20.
Update: Friday 11:30, The nodes are now up again.
During this maintenence it is expected that the temperature will rise so much that we will be forced to shutdown the regatta nodes.
The maintenance is expected to take place from 08:00 to 10:00 tomorrow, Friday 20.
Update: Friday 11:30, The nodes are now up again.