We are having problems with defining a High Available (HA) remote broker setup
We have following configuration.
- JMS Hosts (on a traffic instance)
- Default JMS Host
During startup of an instance â€˜transaction-recoveryâ€™ is performed for JMS
(before EJBs are loaded).
The implementation in Glassfish is such that it tries to make a connection to
all hosts in the â€œJMS Hostsâ€� list (also called â€˜AddressListâ€™).
So a PL will be waiting/blocking (for a certain time) until the broker on the
other PL is ready.
We have tested a scenario where PL_2_4 was shutdown.
During startup of PL_2_3 we see that the PL_2_3 tries to connect to broker
PL_2_4 for about 2.5 minutes (10 retries).
This happens twice because of 2 connection factories for EventQueue and
This means the traffic instance has a (max) delay of about 5 minutes before it
continues (with loading of EJBâ€™s).
As far as can be see there are no configuration parameters to influence this
behavior for â€œtransactions-recoveryâ€�.
Can you please indicate on how to setup a working REMOTE broker setup.
Attachments show more information.