Hi Binti, See below
> On Aug 21, 2016, at 4:24 PM, bintisepaha <binti.sep...@tudor.com> wrote: > > Hi Denis, > we see this exception too from a client when the cluster is > restarted."IllegalStateException: Cache has been closed or destroyed: cache" > we reconnect the client to the cluster by calling Ignition.stop and > Ignition.start/ignite again and we are able to avoid this. > In such a scenario there is no need to restart a client node. You need to get a new reference to a cache using Ignite.cache(…) or Ignite.getOrCreateCache(…) API. This can be done inside of a try-catch block that processes IgniteClientDisconnectedException https://apacheignite.readme.io/docs/clients-vs-servers#client-reconnect <https://apacheignite.readme.io/docs/clients-vs-servers#client-reconnect> In any case I think that Ignite may handle such situations automatically https://issues.apache.org/jira/browse/IGNITE-3719 <https://issues.apache.org/jira/browse/IGNITE-3719> > As far as cluster hanging goes, We are seeing many issues. optimistic txns > hanging in commit(), if we try to kill nodes with hanging txns, the cluster > hangs afterwards. unfortunately we cannot stop using Ignite at this point, > we are already in production with some functionality. > > What can we send you to help us solve this issue? Share logs and thread dumps from all the nodes. — Denis > > Thanks, > Binti > > > > -- > View this message in context: > http://apache-ignite-users.70518.x6.nabble.com/One-failing-node-stalling-the-whole-cluster-tp5372p7199.html > Sent from the Apache Ignite Users mailing list archive at Nabble.com.