Cinder backend should reconnect after failure

Registered by Ivan Kolodyazhny on 2016-05-24

If cinder-volume failed to initialization failed during startup it will never try to re-initialize again. To get it working again operator should restart cinder-volume service.

In some cases, like network outage during startup, it would be useful to have options to allow cinder to re-initialize drivers without restart. E.g.
[driver-conf-section]
reconnect_on_failure=True
reconnect_on_failure_timeout=30

reconnect_on_failure should be disabled by default

Blueprint information

Status:
Not started
Approver:
Sean McGinnis
Priority:
Low
Drafter:
Ivan Kolodyazhny
Direction:
Approved
Assignee:
Ivan Kolodyazhny
Definition:
Approved
Series goal:
Accepted for stein
Implementation:
Unknown
Milestone target:
milestone icon stein-3

Related branches

Sprints

Whiteboard

Gerrit topic: https://review.openstack.org/#q,topic:bp/driver-initialization-after-fail,n,z

Addressed by: https://review.openstack.org/599505
    Driver reinitialization after failure

Addressed by: https://review.openstack.org/618702
    This is a proposal to reinitialize a volume driver when it fails during starting.

Addressed by: https://review.openstack.org/622697
    Add an new exception for environment error

Addressed by: https://review.openstack.org/624853
    nitialization after failure

(?)

Work Items

This blueprint contains Public information 
Everyone can see this information.

Subscribers

No subscribers.