Artemis - strange ack failure when using STOMP

[nkhasanov <nk...@gmail.com>]

Hi everyone,

I have a strange behaviour of Artemis which leads to a situation when broker
stops sending messages out and we have to reboot the service in order to
kick it out. 

I use STOMP protocol. Usually problem appears shortly (several minutes
within an hour) after broker restart, but not immediately and not every
time. First minutes Artemis operates normally, but then at some point
Artemis stops accepting ACKs from clients and in case of receipt it replies
with an ERROR "failed to ack because no message with id: ...".

From the heap dump I see that message_id mentioned in the error above is in
the StompSession.messagesToAck Map, so it's not clear how the situation can
even happen, because message_id is put into the map right before sending the
actual stomp frame to the client.

Did anyone experience similar issues, any ideas in which conditions can it
happen? Maybe you have I'm happy to contribute with the fix, but not sure
even how to reproduce it properly. It happens from time to time, roughly
every 20th restart may lead to that.

I'll investigate on my side with the environment I have, just wanna get some
other inputs and ideas.



--
View this message in context: http://activemq.2283324.n4.nabble.com/Artemis-strange-ack-failure-when-using-STOMP-tp4727490.html
Sent from the ActiveMQ - User mailing list archive at Nabble.com.