Time Synchronization

For best results, you are strongly recommended to synchronize the clocks on all the machines hosting your production replication group. Running a time synchronization daemon like NTPD is a simple way to keep time synchronized across your replication machines. Once the clocks are set, they are maintained by ntpd so that they rarely stray more than 128ms away from one another.

Be aware the JE checks for clock skew between the Master and a starting Replica node, when the Replica node performs its startup handshake with the Master. (See Replica Startup for information on the startup handshake.) If the clock skew between the two nodes is too large, the handshake is aborted and JE throws an EnvironmentFailureException.

Also, well-synchronized clocks are required for a proper implementation of a time consistency policy (see Time Consistency Policies). It is also required for correct internal booking by JE.

Finally, synchronized system clocks make it easier to correlate events in the logging output from different nodes in the group.