heartbeat connection timeout

[Angela Maczka]

What is heartbeat connection timeout and how is its value determined?

Thanks,

Angela

[Stuart Abramson_2]

It's defined in the "host.ascii" file:
CLUSTER_NAME gbo538-d.cluster
NODE_NAME gbo538-d
NETWORK_INTERFACE lan0
HEARTBEAT_IP 151.208.28.138
NETWORK_INTERFACE lan2
NETWORK_INTERFACE lan4
HEARTBEAT_IP 192.168.28.138
HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 2000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
MAX_CONFIGURED_PACKAGES 0

Now, if your question is "what should I set it to?", then you have to read the man pages, and the MC/SG manual (which is very good.). If you have a good LAN, then you don't need to change it, but some people lengthen it.

[doug mielke]

A slow or busy network may require lengthening the timeout, as can a busy or heavily loaded server, even on a clean network.

The heartbeat is basically one system asking the other, 'are you still there' over and over.
timeout determins how long the server should wait for an answer before making the determination that there is a problem with the other server.

[Angela Maczka]

When I do a cmscancl it lists the heartbeat interval and the node timeout as specified in the cluster.ascii file. However it also lists the "heartbeat connection timeout" value. I don't see anywhere this is specified in the cluster file. What is this used for and how it its value derived? Is it calculated based on the other values?

Thanks,

Angela

[Kent Ostby]

Angela --

This value is always set to 2*node_timeout.

It can only be changed in this way.

It is used internally by Serviceguard during the reformation process.

Best regards,

Kent M. Ostby