Cinder

  1. Bug #1536248
  2. Comment #3

Comment 3 for bug 1536248

Revision history for this message
Mitsuhiro Tanino (mitsuhiro-tanino) wrote :

From my further investigation, this is not a bug but we need to know recovery steps for c-vol node crash with lioadm.

From the commit massage in https://github.com/openstack/cinder/commit/706878deaaa5acbb8a577942a5a774c58fe16332,
I think RHEL has a "target.service" and this service restore /etc/targetcli/saveconfig.json after rebooting the server. In this case, I guess LIO iSCSI configuration will be restored automatically.

If you use non RHEL OS, we need to restore existing LIO configuration by hand(or add this to boot script).
Cinder has a function to save LIO iSCSI configuration automatically to the /etc/targetcli/saveconfig.json by default.
We can use this.

So right recovery steps after c-vol node crash are;
(1) Reboot the server
(2) Restore /etc/targetcli/saveconfig.json
       sudo targetctl restore /etc/targetcli/saveconfig.json
(3) Start c-vol service

And then, iSCSI connection from iscsi client will be recovered.