Problem in Lan-Card

[Amit Manna_6]

Hi All,
We have two N-Class servers in a cluster using MC/SG.We have configured redundant Lan-Card for one network segment
The parameters set in the cluster configuration are send as an attachment.


Today morning one of the network segment failed and the segment got switched to the redundant card. And after three hours the segment reverted back to the original card. Following error was reflected in the syslog file.

cmcld: lan3 failed
cmcld: Subnet 137.88.0.0 switched from lan3 to lan5
cmcld: lan3 switched to lan5

There was no error displayed in any of the logs indicating a netwrk failure or card failure.

This occurence happened two times within an interval of 13 hours.

Anybody pls help.

[viks]

Pls set ur heartbeat interval to 1 sec and then observe the behaviour.the recommended setting for heartbeat interval is 1 sec.
try out !!
viks

[Amit Manna_6]

back to top

[Sanjay_6]

Hi amit,

This is the cluster conf info from one of my clusters,

HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 5000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000

Parameters are in microseconds.

Hope this helps.

Regds

[Sridhar Bhaskarla]

Hi Amit,

I keep the following to 8 sec. The default 2 sec is short.

NETWORK_POLLING_INTERVAL 8000000

I do the same for NODE_TIMEOUT too.

Try it.

-Sri