Cinder iSCSI configuration lost on target.service start/restart
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Cinder |
Fix Released
|
High
|
Gorka Eguileor |
Bug Description
This bug relates to iSCSI lio helper.
When target.service is started or restarted all lio iSCSI configuration is lost.
How to reproduce it:
Create an instance and attach an iSCSI (with lioadm as helper) volume to it, the instance can see the volume.
Start or Restart target.service on the server serving the volume, the instance loses connectivity.
Output example:
-Expected results:
[root@example ~]# cinder-rtstool get-targets
iqn.2010-
[root@example ~]# systemctl restart target
[root@example ~]# cinder-rtstool get-targets
iqn.2010-
- Actual results:
[root@example ~]# cinder-rtstool get-targets
iqn.2010-
[root@example ~]# systemctl restart target
[root@example ~]# cinder-rtstool get-targets
Additional info:
This bug happens because rtstool is only making changes to configfs and not making these changes as persistent. So when target.service starts/restarts and looks for desired configuration he doesn't find latest Cinder changes and loads whatever configuration was saved, which usually is an empty configuration.
Desired behavior would be for Cinder to make changes as persistent so there's a coordinated access from multiple APIs (target daemon, Cinder, targetcli...).
Changed in cinder: | |
assignee: | nobody → Gorka Eguileor (gorka) |
Changed in cinder: | |
importance: | Undecided → High |
tags: | added: drivers lvm |
Changed in cinder: | |
milestone: | none → kilo-rc1 |
Changed in cinder: | |
status: | Fix Committed → Fix Released |
Changed in cinder: | |
milestone: | kilo-rc1 → 2015.1.0 |
Fix proposed to branch: master /review. openstack. org/160428
Review: https:/