Starting or restarting an instance of a clone resource that has a Mandatory order constraint causes clones throughout the cluster to restart in a RHEL 6 or 7 High Availability cluster with pacemaker
Issue
- I have two clone sets and a Mandatory order constraint tying them together. When I
standbynode 1, the resources stop on node 1 and keep running on node 2. When Iunstandbynode 1, the clones of the "second" resource in the order constraint restart everywhere throughout the cluster. - When we go to place a node back into the pcs cluster by running
pcs cluster unstandby node, PCS will stop every single openstack service on all 3 control nodes thus causing a complete outage. - if we
pcs resource ban rabbitmqon one node and then clear it it will trigger the cascade of restarts.
Environment
- Red Hat Enterprise Linux (RHEL) 6 or 7 with the High Availability Add On
pacemaker-
Two resource clone sets in the configuration
- These resources are connected through a order constraint. For example, in
pcs resource constraint show --full:
Ordering Constraints: start hascript-clone then start hascript2-clone (Mandatory) (id:order-hascript-clone-hascript2-clone-Mandatory) Colocation Constraints:- The order constraint is of
kind=Mandatory - The "second" resource in the constraint has
interleave=false(or unspecified)
- These resources are connected through a order constraint. For example, in
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.
