Corosync: A processor failed...

[iamspartacus]

I'm seeing these messages very often in my syslog as I'm troubleshooting some unexpected node reboots (a few times a day). Is this being caused by the network dropping out?

Mar 16 14:17:02 athens corosync[6092]:   [KNET  ] link: host: 1 link: 0 is down
Mar 16 14:17:02 athens corosync[6092]:   [KNET  ] host: host: 1 (passive) best link: 0 (pri: 1)
Mar 16 14:17:02 athens corosync[6092]:   [KNET  ] host: host: 1 has no active links
Mar 16 14:17:03 athens corosync[6092]:   [TOTEM ] Token has not been received in 2250 ms
Mar 16 14:17:04 athens corosync[6092]:   [KNET  ] rx: host: 1 link: 0 is up
Mar 16 14:17:04 athens corosync[6092]:   [KNET  ] host: host: 1 (passive) best link: 0 (pri: 1)
Mar 16 14:17:04 athens corosync[6092]:   [TOTEM ] A processor failed, forming new configuration: token timed out (3000ms), waiting 3600ms for consensus.
Mar 16 14:17:04 athens corosync[6092]:   [QUORUM] Sync members[2]: 1 2
Mar 16 14:17:04 athens corosync[6092]:   [TOTEM ] A new membership (1.14b9) was formed. Members
Mar 16 14:17:04 athens corosync[6092]:   [QUORUM] Members[2]: 1 2
Mar 16 14:17:04 athens corosync[6092]:   [MAIN  ] Completed service synchronization, ready to provide service.

For reference I'm using a bond of two 1GbE ports on this host. All traffic is routed over these ports.

 

[fiona]

Hi,
how many nodes are in your cluster? Are you using HA? Corosync traffic needs a low latency, so ideally it's running on a dedicated network. If you are using HA, see here for more information.