Operating the cluster
When running PhenixID authentication services with a clustered configuration it is important that services, nodes and external dependencies are operated, maintained and monitored closely. In general PAS has an central role in the day to day work of the users and should be handled accordingly.
Ensure ports required are opened at all times and that they are not closed by firewalls. Latency between cluster nodes should be kept as low as possible. Since PAS operates in an "eventually consistent" mode, the lower latency the better.
Shorter network glitches should not ba a problem but remaining interference in communications will cause significant performance problems and in some cases complete failure.
It is absolutely essential that all nodes are synchronizing time against the same time sync server(s)
Restarting of services or the entire node should not be done in "un-monitored"manor. In cases where nodes need to be restarted due to OS updates etc. ensure that the nodes have time to reconnected to the cluster before restarting other nodes.