On the same environment as the one described in CONTROLLER-1756 but the problem described here is been seen on that environment but without any OOM I'm told - thus filing this as a separate new issue:

We're seeing ODL "commit suicide" (shut itself down, by System.exit) with the last line in the log being:

2017-08-24 10:03:04,487 | ERROR | -dispatcher-5689 | ActorSystemImpl | 186 - com.typesafe.akka.slf4j - 2.4.18 | Uncaught error from thread [opendaylight-cluster-data-shard-dispatcher-88] shutting down JVM since 'akka.jvm-exit-on-fatal-error' is enabled

Before that there are numerous errors from (...)rd-dispatcher-88 like this:

2017-08-24 10:03:03,446 | ERROR | rd-dispatcher-88 | Shard | 204 - org.opendaylight.controller.sal-clustering-commons - 1.5.2.Carbon | member-1-shard-default-operational, An exception occurred while committing transaction member-2-datastore-operational-fe-1-txn-10344-0
java.lang.IllegalStateException: Store tree org.opendaylight.yangtools.yang.data.api.schema.tree.spi.LazyContainerNode@3cde34f3 and candidate base org.opendaylight.yangtools.yang.data.api.schema.tree.spi.LazyContainerNode@129ff94a differ.

Full log is attached.