HTC Gluster outage planned for Thursday, April 15

[chtc-users@xxxxxxxxxxx]

For users of the HTC System's Gluster filesystem:

In order to correct some recent issues with the quota mechanism for the HTC System's Gluster filesytem, CHTC staff will be taking it down on Thursday afternoon of this week, intending to have it back up by EOB the same day.

To prepare for the outage, we have configured HTCondor to stop matching new Gluster-requiring jobs as of roughly 12pm, today. This may impact your jobs in any of several ways:

• Gluster-dependent jobs that began running before 12pm today will continue running until Thursday afternoon, when we take Gluster down. This way, jobs running within the 72-hour default limit will have time to complete, and no new jobs will be started that might otherwise be interrupted by the downtime. (If any your running jobs will run longer than through 12pm on Thursday, you may want to hold or remove them so that they will not fail at the start of the downtime, and so that they are easy to account for and re-run.)Â
• Gluster-dependent jobs that have not started by 12pm today will remain 'Idle' in the queue and will automatically begin matching once we have Gluster fully operational at the end of the downtime. While you do not need to wait until after the downtime to submit Gluster-dependent jobs, such newly queued jobs will similarly remain 'Idle' in the queue until the end of the downtime.

We will email on Thursday when the Gluster filesystem is up again after the downtime. Please write to chtc@xxxxxxxxxxx with any questions. We appreciate your patience while we correct the issue.

Best,

Your CHTC Team