Github user gemmellr commented on the issue:
https://github.com/apache/activemq-artemis/pull/1534
> "Connection readTimeout has occurred.",
NETTY_READ_TIMEOUT = "read-timeout-seconds";
private int readTimeout;
Would a more descriptive name be better here for the configuration etc,
something that conveys its a one-time protocol header timeout at connection
startup? Using simply 'read timeout' seems like it could be confused with being
related to other more general ongoing TCP read timeout handling and
heartbeating etc, whereas it actually wont have effect beyond the first 8
bytes. Maybe 'protocol-header-timeout' or something like that?
The test was added in the amqp test package as opposed to somewhere more
general, which would seem nicer given it is entirely protocol-agnostic
behaviour under test. Overlooking that though it was added to
"AmqpSendReceiveTest" which seems odd given it obviously doesn't establish an
AMQP connection and send or receive anything. Adding it there also means the
test had to start->stop->reconfigure->restart the broker to perform the
checking. Adding a new test class would be nicer, preferably outside the amqp
package, but even if shoehorning into an existing amqp test there are perhaps
better options like AmqpProtocolHeaderHandlingTest where it could be taken as a
'doesnt send header' test.
---
