[Linux-cluster] Corosync node disconnects occasionally

[Chris Alexander]

Hi all,

Had a bit of an issue with our test cluster configuration after a network
hardware change (which we unfortunately can't reverse easily) and I wanted
to make sure there isn't any known issues with the Cluster software before
we go though another large hardware swapout.

We changed the switching hardware which the cluster software uses for its
administrative data, and now we are occasionally receiving the following
Corosync notification before the node in question gets fenced:

DISMAN-EVENT-MIB::sysUpTimeInstance = <uptime>, SNMPv2-MIB::snmpTrapOID.0 =
COROSYNC-MIB::corosyncNoticesNodeStatus,
COROSYNC-MIB::corosyncObjectsNodeName.0 = "<nodename>",
COROSYNC-MIB::corosyncObjectsNodeID.0 = 1,
COROSYNC-MIB::corosyncObjectsNodeAddress.0 = "<local IP>",
COROSYNC-MIB::corosyncObjectsNodeStatus.0 = "left"

We suspect that the switch is misconfigured or just broken (offsite hardware
hence the difficulties swapping mentioned above) but wanted to check nobody
has had something like this before which might be software related rather
than hardware.

Cheers for any ideas

Chris
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20111029/5481c7e5/attachment.htm>