Cluster with hearthbeat fault
Hi,
I have 2 servers HP DL 385 with RHEL 4 update 4 in clustering with MSA1000. I have Cluster suite 4 ang GFS 6.1. I configured the cluster with a virtual IP, mount FS EXT3 (inside the MSA1000) and a Oracle service. It works fine but I have a problem when I have a fault on hearthbeat channel (Point-to-Point ethernet from server 1 and server 2).
In this case of fault, the servers go in an unpredictable state, in some cases both server 1 and server 2 take the service (IP, mount FS and Oracle). I think that the quorum disk doesn't work very well.
This is what I want if I have hearthbeat channel fault:
- server 2 goes down in a clean mode (ex. shutdown -h now) and not using fence device (ex. ILO)
- server 1 take the service (IP, mount FS and start Oracle)
Anyone can help me ? Or have the same problem ?
Regards
|