[ClusterLabs] Not all bind address belong to the same IP family - corosync fails

[Jan Friesse]

lejeczek,

> hi guys,
> 
> I have a node which fails to start, but only at sys boot time to make it
> more curious, like this:

Config would be helpful, but reason is probably usage of DNS and RRP 
config. Together with delay in DHCP (or DHCPv6) this may cause problem. 
You can ether change corosync.conf to use IPs or add required names in 
the /etc/hosts file (what is recommended practice anyway).

> 
> ....
> 
> corosync[28752]:  [MAIN  ] parse error in config: Not all bind address
> belong to the same IP family
> May 10 12:33:45 rider.privatecorosync[28752]:  [MAIN  ] Corosync Cluster
> Engine exiting with status 8 at main.c:1406.
> May 10 12:33:45 rider.privatecorosync[28530]: Starting Corosync Cluster
> Engine (corosync): [FAILED]
> May 10 12:33:45 rider.privatesystemd[1]: corosync.service: control
> process exited, code=exited status=1
> May 10 12:33:45 rider.privatesystemd[1]: Failed to start Corosync
> Cluster Engine.
> May 10 12:33:45 rider.privatesystemd[1]: Dependency failed for Pacemaker
> High Availability Cluster Manager.
> May 10 12:33:45 rider.privatesystemd[1]: Job pacemaker.service/start
> failed with result 'dependency'.
> May 10 12:33:45 rider.privatesystemd[1]: Unit corosync.service entered
> failed state.
> May 10 12:33:45 rider.privatesystemd[1]: corosync.service failed.
> 
> Would anybody care to suggest how to investigate & fix this?
> 
> many thanks, L.
> 

Regards,
   Honza


> 
> 
> _______________________________________________
> Manage your subscription:
> https://lists.clusterlabs.org/mailman/listinfo/users
> 
> ClusterLabs home: https://www.clusterlabs.org/
>