HPC2N
High Performance Computing Center North
The batch nodes of Abisko is currently down due to a major power failure in Umeå.
The lightning looks to have caused a short-circuit in one of the circuit breakers feeding Abisko.
(To give the electricans a better chance to solve the problem, all power feeds are kept off for the time being, so Abisko is currently completely without power.)
We are currently waiting for electricians to arrive on site.
We are experiencing some problem with /pfs. We are working on it and hope to get it back online soon.
During this problem we have stopped the batch queue (no new jobs will start). As soon as pfs is back we will start up the queue again.
*UPDATE 2016-05-09 16:30*
Pfs back online. Batch queue started again.
Mon, 2016-05-09 13:35 | Roger Oskarsson
Our long awaited upgrade of the Lustre file system to double the size and performance is finally approaching.
On May 2:nd (2016-05-02) Abisko will be down (no jobs running) and the login node will be taken offline.
This is necessary to be able to add the new hardware and do the required recabling/rearranging of the system.
The down time will be at least two days long.
Mon, 2016-04-25 15:21 | Åke Sandgren
Abisko will be down due to maintenance on the cooling system 20160413 and 20160414.
During that time we will also perform some maintenance on the /pfs/nobackup file system.
This means that /pfs/nobackup will be unavailable both of those days.
No jobs will be allowed to run during this maintenance window.
*UPDATE 2016-04-14 19:00*
The system is now back online again.
Mon, 2016-04-04 08:39 | Åke Sandgren
Abisko is now back online after a two day maintenance window.
We have found some discrepancies regarding the quota information that are not yet fixed.
Some users have large differences between what the quota system thinks and what is actually there in the file system. This may cause some problems with hitting the file quota limit. If this happens please send a mail to support@hpc2n.umu.se and we will deal with it.
We will try to fix this problem during the next maintenance window in May.
For unclear reasons, accesses to AFS on the abisko login node just hung. With no way to fix it, we choose to reboot it.
Thu, 2016-03-31 17:47 | Roger Oskarsson
The /pfs/nobackup file system is currently unavailable.
It may have gotten into problems due to the network problem on Abisko.
We are currently trying to bring it back online.
*UPDATE 2016-03-10 14:15*
The file system is now back online and updated to a version that have fixes for the problem we where hit by.
Wed, 2016-03-09 20:34 | Åke Sandgren
Abisko is down due to network problems.
We are investigating the problem.
All queues have been stopped.
*UPDATE 2016-03-09 19:18*
The main ethernet switch for Abisko has died.
We will chase our supplier to fix this asap.
It will not happen until earliest tomorrow (2016-03-10)
*UPDATE 2016-03-09 20:00*
We found a spare power module for the switch and Abiskos ethernet network is now back onlnie.
We will verify system functionality (see other system news about /pfs/nobackup) before bringing the batch queues back on.
Akka has served us well for eight years but will now finally be retired.
Akka will be turned off at the end of March and dismantled to make room for our new upcoming system.
The /pfs/nobackup filesystem has stopped responding.
*UPDATE 2015-12-16 18:10*
The file system is now back online again and the queues are running.
Wed, 2015-12-16 15:21 | Birgitte Brydsö
