Uppsala Multidisciplinary Center for Advanced Computational Science

Problems with Irma's storage system Lupus -- PREPARING FOR A REPAIR


SInce Monday evening Lupus have had problems, giving problems also for logins and jobs on Irma.

UPPMAX is busy trying to solve these problems, and will add more information here when available.

Update 12:56

Things should be working since a while back.

Update at 1320 hours

We are probably getting support from the vendor on the problem later today, but think that this will not disturb your work.

Some Irma compute nodes stopped working. We will now bring them back in production.

Update at 1400 hours

We have again some Lupus problem, or possibly the same again.

Now busy investigating and mitigating.

Update on Sunday at 1500 hours

Hopefully Lupus works again for you. Please send details to our support address with details of any wriing or reading problems you experience with Lupus.

During next maintenance window, Wednesday November 2nd, a part of the Lustre file system will be moved from secondary Lupus server to primary Lupus server. There are also some error messages that are still not explained.

Update on Friday 2016-11-04 at 1730 hours

It is very difficult to use Lupus for writing since the last weekend. It behaves as if it is i read-only mode. UPPMAX has error reported this to our support vendors.

Some users have asked for us to wait until they have made copies of important data, before we repair Lupus, so we are waiting for that, and helping with the copying.

Update on Saturday 2016-11-12 at 1310 hours

Most copies of important data has finished, so we plan to begin the repair on Monday 2016-11-14. Part of the repair is a file system check on two parts of the file system. If unfortunate, these file system checks may not finish for several days, so please have some more patience.

Old System News