Cinder Volume Issues
Created | Jun 24, 2026 |
|---|---|
Type | User Action |
Owner | @Ramzi Jalili |
Summary | Cinder volumes becoming inaccessible or causing VMs to not boot due to volumes being locked on the back end |
Status | Active |
User Action Required | Yes |
Last Updated | Jun 25, 2026 |
Archive Date | Jul 25, 2026 |
Stuck Cinder volumes issue
Summary:
Many users may be experiencing some issues with Cinder volumes after the power blips on Jun 18, 2026 If you:
have trouble mounting a volume
have trouble writing/accessing an exiting mounted volume
have trouble accessing a VM with a mounted volume
have trouble deleting a volume
have trouble creating volume snapshots
Please consider submitting a ticket to cloud-support@stfc.ac.uk
Why?
The power incident on Jun 18, 2026 has left stale locks on ceph RBD volumes that cinder uses. This has resulted in some volumes being stuck in Read-Only mode - preventing writes and making the volume inaccessible. The data on the volume is likely consistent and healthy and has a high chance of being recovered once we remove these stale locks
Depending on how the volume is used - this could affect the boot sequence of the VM - resulting in some VMs with cinder volumes becoming inaccessible.
The Read-Only also results in some cinder volumes are failing to delete or take snapshots
The Fix
To fix this issue, we need to perform a procedure on the ceph side to remove stale locks.
To recover the affected volumes, some backend Ceph maintenance is needed, and this work requires the affected VMs and K8s clusters to be powered off and the Cinder volumes to be detached while this takes place to prevent any data corruption.
If you believe you have Cinder volumes that are affected, please open a ticket to cloud-support@stfc.ac.uk
We will schedule a time with the Ceph team every week to complete this procedure on a batch of volumes at a time.
Please include:
The volume Name and ID
If you want us to recover these volumes - please either:
Shutoff your VMs/Clusters and detach volumes yourself (we’ll give you a time-frame we them off for when you submit the ticket)
Authorize us to shutoff your VMs/Clusters and detach your volumes before we begin the procedure
This will likely take a full day and we will inform you once the procedure is completed so you can restart your clusters/VMs
Please raise a ticket to cloud-support@stfc.ac.uk if you have any questions
Who is affected:
Any users of the Cinder Block Storage service with existing volumes
New cinder volumes - created after the power incident Jun 18, 2026 seem unaffected
Contact
Cloud Support: cloud-support@stfc.ac.uk
Slack: #user-chat