OpenStack Cinder Volumes Become Read-Only After cinder-volume Container Restart
Applies to
- OpenStack environments using Cinder with NetApp ONTAP NFS backend
- ONTAP 9
- Cinder volumes hosted on NFS
- Cinder backends with nas_secure_file_operations and nas_secure_file_permissions parameters
Issue
After restarting the cinder-volume container, multiple Cinder volumes unexpectedly switched to read-only mode, causing service outages for virtual machines (VMs) and The following errors were observed in logs:
ERROR:__main__:Failed to change ownership of /var/lib/cinder/mnt/<volume>/.snapshot/<snapshot> to 42407:42407Traceback (most recent call last): File "/usr/local/bin/kolla_set_configs", line 352, in set_perms os.chown(path, uid, gid)OSError: [Errno 30] Read-only filesystem: '/var/lib/cinder/mnt/<volume>/.snapshot/<snapshot>'
Symptoms included:
- cinder-volume container repeatedly crashing and restarting
- Cinder volumes mounted read-only on VMs, resulting in loss of write access and service outages
- The issue was triggered after a hostname change and subsequent service restarts
