We pushed back a bit on the schedule previously announced. We’ve got compute001 slated for 9AM tomorrow, Wednesday 10/5, and next week we’ll finish off with stomp. As before, during each rebuild, the machine will be unavailable for some portion of that day. We’ll announce the shutdown on the machine itself to all logged-in users …

First up, the Gitlab maintenance announced earlier today is complete. The UI is a bit different – seems to be a more unified interface between desktop/mobile, so if you don’t see what you expect, hit the “hamburger menu” up in the upper left corner. Secondly, CIS has proposed the following potential maintenance weekends: November 4-6, …

In prepping for the previously announced file server upgrade (we need to do it right, and smoothly), we discovered a couple of gotchas. When we scheduled it, I set this afternoon as a go/no-go decision point, and if we weren’t ready to do it today, there’s no way to reliably say we’d be ready to …

We need to apply some updates and reboot the servers xgitlab.cels.anl.gov and gitlab.cels.anl.gov. This will happen at 4PM today. The outage shouldn’t take too long, but we’re scheduling an hour just in case. During this time, you won’t be able to visit or use repositories hosted there. Let me know if this poses an undue …

We’ve scheduled this Saturday, 10/1 from 9AM onward (all day) to finally migrate the rest of our unix homes off the aging file server onto our new one. To do this, we need to take the home filesystem offline. This will not affect e-mail, web, or other services beyond the MCS linux computing infrastructure (https://wiki.mcs.anl.gov/IT/index.php/Linux). …

We’re still seeing issues with our primary file server, and it’s resulting in severe degradation of service at this point, to where logins aren’t being handled. We’re going to be rebooting it shortly. Some machines may require a reboot after this is done. Stand by for updates.

The previously announced outage should be resolved.

Planned power work in the data center took out our file server (which should have stayed up due to redundancy, but we think we may have a bad power supply). The team’s working to bring it back as we speak. Thanks for your patience.

vanquish has a long-running job on it, so we will postpone its rebuild until we can work with the owner of the job. crank has suffered a disk error, so its rebuild is being accelerated to this week (today). Sorry for any inconvenience, and thanks for your patience.