Hi all,

I created two node cluster using redhat cluster suite. I am using a quorum disk as a third member of cluster.

For quorum disk purposes I am using small LUN device created on disk array which is connected to nodes via SAN.

The OS environment is RHEL 5.4 and I am using multipathd software for fibre chanel paths management.


Quorum disk disk is working and is a proper member of cluster but sometimes CMAN lost contact with him. After some time the quorum disk availability is restored and everything back to normal.

Mar 29 12:49:38 volovec openais[21345]: [CMAN ] lost contact with quorum device
Mar 29 12:50:50 volovec openais[21345]: [CMAN ] lost contact with quorum device
Mar 29 12:52:02 volovec openais[21345]: [CMAN ] lost contact with quorum device
Mar 29 12:53:14 volovec openais[21345]: [CMAN ] lost contact with quorum device

This situation is same on both nodes. All members have 1 vote and cluster needs at least 2 votes to start providing resources. When the nodes are up and running the temporary unavailability of quorum disk doesn't have impact to cluster.
BUT when one node is down unavailability of quorun disk cause stopping of cluster resources which is critical problem for me.

I think that issue could be caused by slow responces of quorum disk device. Maybe I need to change some cman parameter in cluster.conf file but I am not an expert on this.

My initial configuration was:

<quorumd device="/dev/dm-3" interval="12" min_score="1" tko="5" votes="1"/>
<totem consensus="4800" join="60" token="10000" token_retransmits_before_loss_const="20"/>

I try to change the parameters to following values without any sucess:

<quorumd device="/dev/dm-3" interval="12" min_score="1" tko="10" votes="1"/>
<totem consensus="9600" join="120" token="20000" token_retransmits_before_loss_const="40"/>

Please could you help me to find out what to do? I am doing something wrong or which CMAN parameter has to be increased/decreased to resolve my issue.

thank you very much.

regards,
Stefan