-
1. Re: Async error occurred Can not remove a consumer from a session not regd
garytully Mar 11, 2011 6:42 AM (in response to satyab7)A full log of the broker and of the client(With trace level loggin) would help here. It will be possible to correlate requests with the shutdown process.
It looks very much like an ordering issue with a remove command. The failover transport replays what ever command was in progress when the connection to the broker is recreated. If the broker is also in the process of tearing down the connection, the it can get there first and remove the session etc such that the replayed remove command is ignored (as in this case)
-
2. Re: Async error occurred Can not remove a consumer from a session not regd
satyab7 Mar 11, 2011 12:33 PM (in response to garytully)Attached is activemqlog at trace level as requested.
This warning msg can be seen even if I just shut down the client (jbosserver) or restart.
Also can you please elaborate on your below comments.
Can I do some configuration settings in failover URL of client that help with this problem
COMMENTS : "It looks very much like an ordering issue with a remove command. The failover transport replays what ever command was in progress when the connection to the broker is recreated. If the broker is also in the process of tearing down the connection, the it can get there first and remove the session etc such that the replayed remove command is ignored (as in this case)"
the following is failover URL. One thing I have struggled is how to pass multiple transport options to failover URL say I want to pass maxReconnectivityInterval and timeout , I could not. I was hoping by doing that I could probably control these warning msgs.
Truly appreciate your help.
-
activemqlog.txt 1,023.7 KB
-
-
3. Re: Async error occurred Can not remove a consumer from a session not regd
satyab7 Mar 16, 2011 4:48 PM (in response to satyab7)Thanks.
From your other reply I could make multi parameters work in failover URI. But passing
maxReconnectDelay=30000
initialReconnectDelay=10000
did not help solve this problem.
I am assuming it is Jencks connection factory that is closing connections or 5.4.2 broker that tries to remove consumers from closed connections , could not figure out. But in my testing I noticed this message is also there even in 4.1.1 but just a debug message.
Where as in 5.4.2 it is elevated to WARN message.
As long as this is just a warning and does not interfere with functionality I do not care.
Spring and Jencks , abstraction did help but cant figure out whats going on in those factory classes.