Planned maintenance on Fram on 16.10.2019

Update:

  • 2019-10-18 14:36 We are ready with the reinstallation, configuration checks, QA and tests. Access to the machine has been reopened and queued jobs are running again.
  • 2019-10-18 06:12 Reinstallation of compute nodes is much slower then anticipated and thus re-opening of the machine is delayed. We do our best to finish the maintenance as soon as possible. In parallel we are conducting tests and benchmarks.
    Will keep you updated.
  • 2019-10-17 08:25 File system servers and infrastructure switches were patched yesterday.
    We are proceeding now with the upgrade of the service and the login nodes.
  • 2019-10-16 08:07 Maintenance has started.

Dear Fram User,

We will have a two days planned downtime starting from 08:00AM on the 16th of October for maintenance on the storage and the file system.

During this time we will, together with the vendor, upgrade the storage firmwares, upgrade the software on the /cluster file system servers and upgrade the operating system on Fram.

This upgrade is necessary to fix the frequent issues with the metadata servers and enhance stability and security of the system.

Fram jobs which can not finish by the 16th of October, are queued up and will not start until the maintenance is finished.

Thank you for your consideration!

Metacenter Operations

Fram machine room cooling problem

Dear Fram cluster users:

We have a problem with the cooling system in the Fram machine room,
due to this, we have to reduce the load on the cluster by reserving the entire cluster, which means no job will run.
We are sorry for the inconvenience, and we will keep you updated.

Update 2019.07.11 08:00: Fram should be fully operational again, we are monitoring the machine and releasing compute nodes back to production.

Update 14:55: Some of the nodes are crashed, which means it’s possible that  some of the jobs get killed

Update 2019.07.03 10:55:  To keep the machine room temperature reasonably low with only one working CDU, we have kept 495 nodes in maintenance state while 197 nodes are in downstate, we will monitor the power consumption in the machine room and release more nodes accordingly.

Update 2019.07.08 11:55:  Fram is expected to be back to full its full capacity on Wednesday, 2019.07.10.

Fram development queue

Dear Fram User,

As of today we have adjusted the queue system policies to facilitate code development and testing on Fram and meanwhile limit possible misuse of devel queue.

devel is now adjusted to allow:

  • max 4 node jobs
  • max 30 minutes wall time
  • max 1 job per user

We have additionally introduced a short queue with following settings:

  • max 10 node jobs
  • max 120 minutes wall time
  • max 2 jobs per user

We will continue to monitor and improve the queue system. Please stay tuned.
You may find more information here.

Metacenter Operations

Fram MDS patched

Dear Fram User,

This morning around 09:05, once again has the Fram metadata server crashed and likely had impact on running jobs.

A mitigating patch was delivered by the vendor yesterday and we used this opportunity to apply it on our metadata servers.

We will keep the system closely monitored and cooperate with the vendor on further stabilizing the system.

Apologies for any inconvenience this may have caused!

Fram MDS crashed

Dear Fram User,

Once again has the Fram metadata server crashed and likely had impact on running jobs.
We are in contact with the storage vendor for patching the file system.

Apologies for the inconvenience!

Clean-up of Fram filesystem needed

Dear Fram users,

The Fram filesystem, and most critically /cluster/work and /cluster/home, is running out of inodes, and there is only 8% left. If we run out of inodes it will not be possible to create new files. To avoid loss of data and job crashes we kindly ask all of you to if possible delete files that you no longer need.

Best regards,

Jon, on behalf of the Operations Team