This is Lite Plone Theme
You are here: Home System Status RESOLVED: CXFS problem(s?)

RESOLVED: CXFS problem(s?)

Dear users,

Yesterday some of you may have noticed a short glitch with /fast/space on universe. Unfortunately, the "short glitch" has come back and it has grown to full adulthood, being now a "big problem": /fast/space is inaccessible all across the systems.

We are investigating this with SGI but cannot yet give any estimate of its return to service. Some jobs will die as a result. However, other filesystems are still up and running although it is yet unclear if we need to reboot the machines to sort this out, so please consider all machines free to use (but not /fast/space) but at-risk until further notice. Further status reports will be posted here.

We are sorry for the inconvenience caused.

Best regards, COSMOS Management

UPDATE@2015-05-14T15:51+0100

Dear users,

After unsuccessfully trying to solve the problem with /fast/space we now need to unmount all CXFS filesystems on all computers. This means all jobs die and quite likely some interactive logins will need to be killed, too. We will try to avoid rebooting any of the machines (apart from universe which crashed during the attempts to keep the problem isolated) but please consider them at extreme risk from now on.

UPDATE@2015-05-14T22:37+0100

Dear users,

Service has been restored, but unfortunately only a single job survived. We are sorry for the inconvenience caused and still investigating the root cause.

Best regards, COSMOS Management

We are sorry for the continued inconvenience.

Best regards, COSMOS Management