Hello. I'm not sure if someone will have an explanation, but I'm going to send my question... in a bottle !
We had a strange behavior last night : our PostgreSQL server has been failing with a raher fatal error, and has switched into recovery mode for some time. Usually, when an SQL error occurs the ActiveMQ brokers log exception, pooled connections are removed, and renewed when the DB server is back. But last night after closing all the failing connections, the brokers (we have 12 on a given environment) have all stopped renewing JDBC connections, and logically refused any client JMS activity : all the system has been freezed waiting for the brokers. Restarting the brokers has solved the issue, but we'd like to know if this behavior of the JDBC persistence adapter is known, normal or not ? Our production version is very old : 5.11 (in red-hat fuse). Perhaps newer versions do not behave the same way... Some expert opinion ? Thanks for your help. Regards.