cinder failed to login in to solidfire backend when creating volume

asked 2020-01-17 05:38:20 -0500

rowbot gravatar image


We are busy deploying OpenStack Stein using kolla-ansible and have hit an issue with cinder and our solidfire backend. Background information that may be useful...

  • 7 compute hosts all with iSCSI configured for swift
  • glance setup to use the swift backend (this is working)

When we create a volume from the UI or command line there is an error reported from the cinder_volume container. Command used:

openstack volume create -vvv --size 35 --description "test image" --image 4accff66-0391-4a6a-a019-d1a11d19e146 test-image

cinder_volume logs of failure:

Failed to login iSCSI target on portal 192.168.x.x:3260 (exit code 12).: oslo_concurrency.processutils.ProcessExecutionError: Unexpected error while running command.
Failed to connect to iSCSI portal 192.168.x.x:3260.
Couldn't find iSCSI nodes because iscsiadm err: iscsiadm: No records found
: os_brick.exception.VolumeDeviceNotFound: Volume device not found at .

The volume then goes into an error state. This error only occurs if we create a volume using the --image flag. Just creating a volume normally does not seem to have any errors.

The image appears to download correctly from glance as I can see this debug in the logs just before the iSCSI error:

Image download 963.62 MB at 74.33 MB/s

Our cinder.conf file does contain the [solidfire] configuration options and we can see that the volume is created on the solidfire backend.

Has anybody encountered this error before or have any ideas how we can get to the root of the problem?

I am not sure where target iqn comes from, I assume that the solidfire driver (cinder.volume.drivers.solidfire.SolidFireDriver) used by cinder is discovering this or generating it?

I also tried running iscsiadm inside the cinder_volume container myself but I am not sure what to expect:

$ iscsiadm -m discovery -t st -p 192.168.x.x
iscsiadm: No portals found
edit retag flag offensive close merge delete