Java CI and DI instance re-starts after an HA failover of the SCS instance.

[Former Member]

Hello,

We have installed a SAP system onto an IBM HACMP cluster with an SCS and corresponding ERS running on seperate nodes in the cluster.

We have tested that the ERS is working correctly during a failover using ensmon and enqt but the CI and DI instances both restart sometime after a failover test.

We have increased the ms.reconnect.timeout and the ms.notification.timeout to try to resolve this but it still re-starts after the failover.

There's a timeout message appearing in the default.trc file in the server log directory but we can't figure out which timeout it is acutally for.

Does anyone know what else we could check/change to try and stop the re-starts?

Regards

Garry Bayford

[Former Member]

Hello,

We've now found that the timeout messages are not really relevent.

The manual HA failover runs a standard stopsap which forces the Java Apps servers to re-start.

We figured a way round this which involves killing the enque process prior to the failover so that no markers can be written to the lock table during the shutdown. So the Apps servers remain running after the manual failover.

Regards

Garry