You are viewing a plain text version of this content. The canonical link for it is here.
Posted to gitbox@activemq.apache.org by GitBox <gi...@apache.org> on 2020/02/09 18:15:55 UTC
[GitHub] [activemq-artemis] franz1981 edited a comment on issue #2975:
ARTEMIS-2617 Improve AMQP Journal loading
franz1981 edited a comment on issue #2975: ARTEMIS-2617 Improve AMQP Journal loading
URL: https://github.com/apache/activemq-artemis/pull/2975#issuecomment-583875806
@michaelandrepearce I suppose you are interested in numbers: in my syntethic tests a 2 GB journal with very small messages (to create tons of garbage due to the infrastructure of messages) wasn't able to be loaded by a broker with `4GB` heap size, always leading to a OOM: now it can load without relevant GC pauses eg
We're not quite yet at the same level of Core messages, but very near: I think that the duplicate `address` is responsible of the difference here.
The commit that allows to half-brute force search `x-opt-delivery-time` is what I'm not sure about:
- that method should be optimized and performed optimally by proton itself IMO (nudge nudge @tabish121 @gemmellr), but I would be super happy to provide a PR for it
- currently it can give rare false positive (causing message scanning when not necessary)
- I've yet to verify how it behave with many message annotations and longer messages
If @michaelandrepearce has some real journal to test with and compare with `master` it would be great, given that this change can affect positively your usage of the broker ;)
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
With regards,
Apache Git Services