CephFS experiencing slow client IO due to full disks Monday 17th February 2025 07:50:00


We're currently experience slow client IO on the CephFS mounted filesystem on ilifu, as part of the process of Ceph balancing data across disks. Users maybe experience this as longer than normal job runtimes as write speeds are currently being throttled. The technical team is working to resolve this issue and restore client IO performance.

We have identified a faulty hard drive in the Ceph storage pool. This disk has now been replaced. The necessary data re-balancing following this hardware change has finished. We are now monitoring Ceph's performance until Monday (21 April) to ensure that the main causes of issues have been resolved.

The CephFS filesystem is currently experiencing degraded write performance. The technical team is working to resolve the issue.

Ceph write performance has returned to normal. The technical team will continue to monitor the situation.