Best way to gracefully shut down a storage enabled node?

snidely_whiplash

What's the best way to shutdown a storage enabled node?  For example if I have backup-count = 0 and I don't want to lose data when reducing cluster size by 1 node.  There does not appear to be a way in the coherence console.    Is there a kill signal that the JVM will honor for this?   Or an MBean operation I should use?

Joefialli-Oracle

One can configure a shutdown hook of "graceful" in coherence operational configuration file. See attached file for an example.

Or one can configure "graceful" shutdown by starting cache server with system property of -Dtangosol.coherence.shutdownhook=graceful.

As long as the kill of the cache server process is not with -9, the configured shutdown hook of graceful will call Cluster.shutdown().

Additional details are documented here for stopping cluster members from command line:

https://docs.oracle.com/cd/E18686_01/coh.37/e18677/cluster_lifecycle.htm#COHDG5473

********

A graceful shutdown can be accomplished via stopping the cache server programmatically calling DefaultCacheServer.shutdown().

Doc link for details: https://docs.oracle.com/cd/E18686_01/coh.37/e18677/cluster_lifecycle.htm#COHDG5474

Harvey Raja-Oracle

You can shutdown the services via JMX or perform a regular kill of the process. The problem with a regular kill is the grace period given to killing the process.

With a graceful termination of the process or shutting down of the service the partitioned service attempts to transfer owned primary partitions out prior to shutdown.

Joefialli-Oracle

One can configure a shutdown hook of "graceful" in coherence operational configuration file. See attached file for an example.

Or one can configure "graceful" shutdown by starting cache server with system property of -Dtangosol.coherence.shutdownhook=graceful.

As long as the kill of the cache server process is not with -9, the configured shutdown hook of graceful will call Cluster.shutdown().

Additional details are documented here for stopping cluster members from command line:

https://docs.oracle.com/cd/E18686_01/coh.37/e18677/cluster_lifecycle.htm#COHDG5473

********

A graceful shutdown can be accomplished via stopping the cache server programmatically calling DefaultCacheServer.shutdown().

Doc link for details: https://docs.oracle.com/cd/E18686_01/coh.37/e18677/cluster_lifecycle.htm#COHDG5474