Large OpenStack Volumes Taking too Long to Delete

Problem

You are trying to delete large OpenStack volumes but the deletions are taking too long. The following log entries are observed in the /var/log/pf9/cindervolume-base.log file.

Javascript

2022-03-23 20:53:28.137 INFO cinder.volume.utils [...] Performing secure delete on volume: /dev/mapper/cinder--volumes-volume--f1846afc--2e87--4ee8--ba2d--b3cfc004d035
2022-03-23 21:15:55.967 INFO cinder.volume.utils [...] Volume copy 307200.00 MB at 227.97 MB/s
2022-03-23 21:15:56.391 INFO cinder.volume.drivers.lvm [...] Successfully deleted volume: f1846afc-2e87-4ee8-ba2d-b3cfc004d035

Environment

• Platform9 Managed OpenStack - All Versions

• Cinder

Cause

The volume_clear parameter is set to zero. This will write a single pass of zeroes to the device before deleting it.

Resolution

1. Edit the /opt/pf9/etc/pf9-cindervolume-base/conf.d/cinder_overrides.conf file and set the volume_clear property to none on all Cinder hosts.

Javascript

[pf9@test-host conf.d]$ cat /opt/pf9/etc/pf9-cindervolume-base/conf.d/cinder_overrides.conf
[DEFAULT]
glance_api_servers = http://10.80.15.19:9292
debug=true

[lvm-backend]
volume_clear = none
target_ip_address = 10.80.16.20
lvm_type = thin

1. Restart the pf9-cindervolume-base service on the Cinder host.

Javascript

# sustemctl restart pf9-cindervolume-base

Additional Information

https://docs.openstack.org/security-guide/block-storage/volume_wiping.html