jms client failure + reconnect
Hello,
I try to achieve the following setup - I have two (or more) queue readers on a Weblogic queue that share the load in the queue
for processing. They are unaware of each other presence.
They all read the message , do some subsequent processing that can take a couple of seconds
and finally do an explicit client acknowledgement.
This works fine in a uninterrupted situation where one sees that they both take part in processing a fair share of the load and all messages are processed.
However killing one of the participants (@work) seems to put the records in progress in some state @the server so that they are not available to the other still
working instance. Even restarting the original instance not always resolves the problem and still another restart is needed.
Could someone please explain the internal logic/rationale behind this (what happens exactly) and in what way the relevant settings (many) need
to be configured to remedy this?.
Currently we experience problems because these "stuck "records" are not delivered in a timely fashion.
Thanks in advance
Peter Geerts