Uppsala Multidisciplinary Center for Advanced Computational Science

Replacing (nearly) all disks on Irma's compute nodes -- DONE

2017-08-23

We have already replaced a lot of disks on Irma's compute nodes, when they got bad, for an unknown reason.

Now we have leared the reason: There is a bug in the firmware of the disks.

But the disks can not be reprogrammed in our computer room.

This means that the vendor will come here to replace the disks, about 860 of them.

The plan is:

  1. UPPMAX will drain half of the nodes for maintenance
  2. The vendor will replace the disks in them, starting on Monday next week
  3. UPPMAX will put those nodes back in production
  4. UPPMAX will drain the other half of the nodes for maintenance
  5. The vendor will replace the disks in them
  6. UPPMAX will put those nodes back in production

In this way, jobs will be able to continue to run all the time.

Update Monday at 1750 hours

More than half of the disks are already replaced.

Update Wednesday at 1350 hours

UPPMAX's support vendor plans to replace disks on the remaining 100 compute nodes on Friday 2017-09-01.

Update Monday 2017-09-04 at 1220 hours

Nearly everything is now finished. Disks of four compute nodes remain to be replaced.

System News