Galera rebuild failed with pacemaker

[lieni]

Detailed bug description:
Miratis 8.0 with Ubuntu
 HA Deployment 3 Nodes Galera, simulate Network outage or Powerloss
 After NEtworking is back online, pacemaker was not able to recover all resources.
 RabbitMQ failed and MYSQL as well.

Steps to reproduce:
 deploy 3 controller node cluster, power off Switches. Rsume Switches

Expected results:
pacemaker should recover resources.

Actual result:
RabbitMQ and Galera MYSQL down.
Failed actions:
    p_mysql_start_0 on sm4.domain.tld 'unknown error' (1): call=764, status=Timed Out, last-rc-change='Tue May 3 19:04:37 2016', queued=0ms, exec=300003ms

PCSD Status:
  192.168.199.3: Offline
  192.168.199.5: Offline
  192.168.199.6: Offline
What means PCSD Status : Offiline, nothing found on this in google.

Workaround:
 Steps we tried to recover:
pcs resource debug-start clone_p_rabbit-server
rabbitmq started on 1. node, on other nodes stopp_app join cluster, start_app
RabbitMQ restored OK.

Galera dosent start up
pcs resource debug-start clone_p_mysql
*** Error in `/usr/sbin/crm_resource': free(): invalid pointer: 0x000000000087a5b0 ***
resource clone_p_mysql is NOT running
resource clone_p_mysql is NOT running
resource clone_p_mysql is NOT running
   error: resources_action_create: A service action must have a valid standard.

Looked for the node with the newest WSREP state.
found one, others have -1 reported.
tried to start that one with pcs cleanup and debug-start
Galera still down.

Ist this a bug or do we just missed something?

[Oleksiy Molchanov (omolchanov)]

Please provide diagnostic snapshot.

[lieni (oliver-lienhard-b)]

Hi
Sorry dont have diag snapshot, deployment was allready deleted.
We will try to reproduce in 1-2 weeks

[Dmitry Pyzhov (dpyzhov)]

Bug is in incomplete state for a month. Closing as invalid. Please reopen if you have more data.