MySQL OCF: pacemaker resource refuses to start after an op start has timed out
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Committed
|
Medium
|
Bogdan Dobrelya |
Bug Description
If operation start timed out, mysql resource refuses to start ever, unless cleanup manually.
How to reproduce:
* Deploy a cluster's as the following:
- DB primary role to deploy as always,
- the rest of the nodes with DB role to start with a modified OCF RA action start to sleep for a period of time higher than op start timeout (default 300)
Expected:
After op start timed out in logs, pacemaker shall repeat start attempts for the resources
Actual:
It times out and remains stopped, until cleaned up manually
Logs snippet:
May 10 11:50:36 n3 ocf-mysql-wss: INFO: p_mysql: get_master_
May 10 11:50:36 n3 ocf-mysql-wss: INFO: p_mysql: check_if_
May 10 11:50:46 n3 ocf-mysql-wss: INFO: p_mysql: check_if_
May 10 11:50:57 n3 ocf-mysql-wss: INFO: p_mysql: check_if_
May 10 11:51:07 n3 ocf-mysql-wss: INFO: p_mysql: check_if_
May 10 11:51:18 n3 ocf-mysql-wss: INFO: p_mysql: check_if_
May 10 11:51:25 n3 lrmd[157]: warning: p_mysql_start_0 process (PID 1220) timed out
May 10 11:51:25 n3 lrmd[157]: warning: p_mysql_
May 10 11:51:31 n3 crmd[160]: notice: Operation p_mysql_monitor_0: not running (node=n3, call=38, rc=7, cib-update=48, confirmed=true)
Solution:
Make MySQL OCF config param master_timeout to be lower than op start timeout
Changed in fuel: | |
importance: | Undecided → Medium |
milestone: | none → 10.0 |
description: | updated |
tags: | added: galera pacemaker |
Fix proposed to branch: master /review. openstack. org/314533
Review: https:/