Bug 1647453 - Pacemaker resource constraints cause API outage during maintenance
Summary: Pacemaker resource constraints cause API outage during maintenance
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat OpenStack
Classification: Red Hat
Component: openstack-tripleo-heat-templates
Version: 12.0 (Pike)
Hardware: Unspecified
OS: Unspecified
high
urgent
Target Milestone: ---
: ---
Assignee: Emilien Macchi
QA Contact: Gurenko Alex
URL:
Whiteboard:
Depends On: 1647449 1647451 1647452 1680280
Blocks:
TreeView+ depends on / blocked
 
Reported: 2018-11-07 13:56 UTC by Lukas Bezdicka
Modified: 2019-04-03 09:38 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of: 1647452
Environment:
Last Closed: 2019-04-03 09:38:27 UTC
Target Upstream Version:
Embargoed:

Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Launchpad 1802108 0 None None None 2018-11-07 13:56:32 UTC

Description Lukas Bezdicka 2018-11-07 13:56:33 UTC 
+++ This bug was initially created as a clone of Bug #1647452 +++

+++ This bug was initially created as a clone of Bug #1647451 +++

+++ This bug was initially created as a clone of Bug #1647449 +++

In more targeted testing of OSP10 minor update we found out there is an issue with way pacemaker services are created and how we defined the order constraints. We will have to provide fix for all the releases of OSP, but for OSP10 we probably will go with workaround I created and tested. Issue is pacemaker resource order constraints are kind Optional which means they do not apply on shutdown of pacemaker cluster on the node. This causes haproxy to be stopped before the VIP is migrated away from the node and subsequent APIs failure. Migration of VIPs will be applied in yum_update.sh script but we should also update the reboot documentation/procedure for the operators.
Comment 1 Michele Baldessari 2019-04-03 09:38:27 UTC 
OSP12 is EOL and we fixed this in the other versions, so closing.

Let me know if you need to discuss this.
Note You need to log in before you can comment on or make changes to this bug.