changing node_timeout on cluster

[p7]

hi all

i have a 2 node cluster and our network is flacky, i see many reformations in the syslog. i want to double the node_timeout in ascii file (curently 10000000). i have an 11.11 dome with 11.15 serviceguard level.

2 questions:

1. what is the proper procedure to do this?

2. with my levels, does anyone see any problems with me doing this?

thx in advance

[melvyn burnard · Accepted solution]

1. what is the proper procedure to do this?Halt the cluster.
Modify the cluster ascii file to reflect the new value.
use cmapplyconf and apply the new value.
start the cluster.



2. with my levels, does anyone see any problems with me doing this?
Well first of all, that means that the cluster will take a LOT longer to react to a failure situation where a node "dies".
Therefore your package failover will be longer. You would be better trying to identify WHY you are seing the reformations and cure taht. Normally causd by high network traffic.

As and aside, Serviceguard A.11.15 has been out of support for sometime, I would STRONGLY advise you upgrade to the last version of Serviceguard to be supported on HP-UX 11.11, which is A.11.16, and also make sure you apply the latest SG patch for that OS/SG version