Reported Outages

New instances failing to provision at CHI@UC

Resolved Posted by Adam Cooper on October 12, 2022
Outage start Wednesday, October 12, 2022 3 p.m.
Expected end Thursday, October 13, 2022 5:32 p.m.

Nodes at UC are currently failing to provision due to an issue communicating with the switches’ control plane, staff are investigating.

CHI@TACC Skylake nodes unavailable

Resolved Posted by Cody Hammock on September 29, 2022
Outage start Wednesday, September 28, 2022 1 p.m.
Expected end Monday, October 17, 2022 7:49 a.m.

Resolved: The failed switch has been replaced, and nodes are once again available.

Due to hardware issues with a swtich, The Skylake compute nodes for CHI@TACC are currently unavailable. Staff are working to restore connectivity for these nodes.

TACC and Authentication Scheduled Maintenance

Resolved Posted by Adam Cooper on September 16, 2022
Outage start Monday, September 26, 2022 11 a.m.
Expected end Monday, September 26, 2022 12 p.m.

The TACC site and our authentication system will be down for scheduled maintenance for an hour on September 26th. 

Authentication issues affecting some users

Resolved Posted by Michael Sherman on September 12, 2022
Outage start Saturday, September 10, 2022 12 p.m.
Expected end Monday, September 12, 2022 6 p.m.

This has been resolved: Users receiving "No active allocations" in error can fix it by logging out and back in.

Chameleon Portal, CHI@TACC, KVM@TACC, and CHI@EDGE networking outage

Resolved Posted by Mark Powers on September 02, 2022
Outage start Friday, September 02, 2022 10:46 a.m.
Expected end Friday, September 02, 2022 3:04 p.m.

UPDATE: All services should now be working as expected

---

Due to a networking issue at TACC, Chameleon’s portal, CHI@TACC, KVM@TACC, and CHI@EDGE are currently unavailable. 
This affects site access as well as already running resources. Site networking staff are investigating, but there is no ETA
for resolution at this time.
 

Object Store at UC currently unavailable

Resolved Posted by Michael Sherman on August 24, 2022
Outage start Wednesday, August 24, 2022 5:22 p.m.
Expected end Thursday, August 25, 2022 10:52 a.m.

The outage is resolved, and the object store is available again.

A rebalancing operation in the underlying storage cluster caused higher than normal I/O, and pushed some of the storage nodes into an unstable state. The node configuration has been tuned to prevent this instability from recurring.

Scheduled downtime for authentication systems

Resolved Posted by Adam Cooper on August 23, 2022
Outage start Tuesday, September 06, 2022 11 a.m.
Expected end Tuesday, September 06, 2022 12 p.m.

On Tuesday, September 6th, our authentication services will be down from 11 AM to 12 PM US Central Time.

http requests being dropped for instances at CHI@UC

Resolved Posted by Michael Sherman on August 18, 2022
Outage start Wednesday, August 17, 2022 12 p.m.
Expected end Thursday, August 18, 2022 3:29 p.m.

3:30PM: The issue has been resolved.


We're observing that http requests to non-https endpoints from nodes at CHI@UC to the internet are being dropped, and are working with our upstream network provider to investigate.

As a workound, you can make requests to an https endpoint, if available.

Jupyterhub storage slowness

Resolved Posted by Michael Sherman on August 05, 2022
Outage start Friday, August 05, 2022 12 p.m.
Expected end Friday, August 05, 2022 2 p.m.

Update 4PM: https://jupyter.chameleoncloud.org/ is back online, and the storage system has stabilized.

Resolving this required more changes than expected, so there may be some remaining instability over the weekend.
Please reach out to the helpdesk if you encounter issues.

Datacenter Cooling failure for CHI@UC

Resolved Posted by Michael Sherman on August 01, 2022
Outage start Monday, August 01, 2022 4:34 p.m.
Expected end Monday, August 01, 2022 9 p.m.

Update 9pm: The chillers have been repaired, and we've received the all clear from ANL staff. CHI@UC is now back to normal operation.

The datacenter hosting CHI@UC has experienced a failure in its cooling system. To reduce load on the remaining cooling, we are blocking new node reservations at UC until the failure is resolved.