Protocol renegotiation does not occur on failover
-------------------------------------------------
Key: QPID-959
URL: https://issues.apache.org/jira/browse/QPID-959
Project: Qpid
Issue Type: Bug
Affects Versions: M3
Reporter: Aidan Skinner
Fix For: M3
The current failover mechanism does not fully take account of the fact that a
session may change protocol versions when failover occurs. Consider:
Broker A is a 0-10 broker
Broker B is a 0-9 broker
Client C is a 0-8, 0-9 and 0-10 capable client
1. C connects to A, with B configured as as it's second broker, with round
robin failover. It negotiates to use 0-10 and begins messaging.
2. Broker A fails and C connects to Broker B. It should renegotiate the
protocol and start talking AMQP 0-9. Messaging should continue as transparently
as possible.
3. Broker B fails and C connects to Broker A. It should renegotiate the
protocol and resume talking AMQP 0-10, messaging should continue.
Currently step 2 will fail
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
