The Rejoin Behavior of a Distributed Cache

The Rejoin parameter enables a cache manager to automatically reconnect to a Terracotta Server Array from which it has become disconnected. This option eliminates the need to manually reinitialize a cache manager to re-establish a broken connection to the Terracotta Server Array.

Integration Server automatically enables the Rejoin parameter for all cache managers whose Terracotta Server Array URLs parameter is set. You cannot disable the Rejoin setting for such cache managers.

Note:
When a cache manager rejoins a Terracotta Server Array, it joins the array as though it were a new client. This means that Integration Server clears any data that it currently holds locally for the cache and resynchronizes its on-heap store with data from the distributed cache on the Terracotta Server Array.

Note:
If your Integration Server becomes disconnected from the Terracotta Server Array frequently, you might encounter OutOfMemory errors. These errors can occur because certain class definitions are reloaded each time the Integration Server rejoins the Terracotta Server Array. If many rejoins occur in a short period of time, disused class definitions can fill the permanent generation space on the heap before they can be garbage collected. For information about this error condition and how you can adjust the size of the permanent generation area to avoid it, see “Avoiding OOME from Multiple Rejoins” in the Terracotta product documentation.

For more information about the rejoin feature of a distributed cache, see “Rejoin” in the BigMemory Max product documentation for 4.1 at www.terracotta.org/documentation.