gemmellr commented on a change in pull request #3887:
URL: https://github.com/apache/activemq-artemis/pull/3887#discussion_r783810708
##########
File path:
artemis-server/src/main/java/org/apache/activemq/artemis/core/remoting/impl/netty/NettyAcceptor.java
##########
@@ -443,6 +443,14 @@ public ActiveMQThreadFactory run() {
logger.debug("Acceptor using native kqueue");
} else {
+ if (useEpoll) {
+ ActiveMQServerLogger.LOGGER.nettyEpollNotAvailable(name);
+ }
+
+ if (useKQueue) {
+ ActiveMQServerLogger.LOGGER.nettyKQueueNotAvailable(name);
+ }
Review comment:
If you are going to have warnings I think youll probably have to check
the env being used just to cut down on the erroneous logs during cases it can
never work, but yes the default xml config should change too.
That said, folks love to copy existing config forward to new installs and so
many might still have it explicitly set even though it doenst apply to them in
their use, especially since it has never really been a 'will use epoll' setting
and is more of a 'can use, in the cases it can work'. So I think thats still
going to lead to enquiries from the folks it incorrectly logs a warning for. On
the flip side it also seems likely many wont set it explicitly themselves for
totally pristine new installs as they either just don't know about it or have
never needed to (and still wont) change it given it defaults to true, meaning
the warnings wouldnt show for them anyway.
Its also fairly typical for folks to have configs that might be used in
multiple different environments which arent necessarily theirs (though having
disparate development and testing/prod environments is also actually pretty
common), so for them they would have to avoid setting it explicitly to avoid
people getting erroneous warnings in the envs its not expected to and cant work
(though adding env checks would reduce those). This is probably more typical on
the client side though, where I'd guess it is less likely to be set explicitly
though (which would mean the warnings wont be seen).
I think it would be simpler all round just to just always log what _is_
being used, at least for the broker where its a one-off event vs
every-connection for the client, where what you propose might be better (though
wont actually be seen unless explicitly toggled).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
