gemmellr commented on PR #6036: URL: https://github.com/apache/activemq-artemis/pull/6036#issuecomment-3505174124
> I understand - most of my CICD experience comes from gitlab (and non-java projects). I don't find it hard to believe a more manual approach works better, I've worked on a couple of my own ... I did however see some outputs of jars being downloaded (on PRs not by dependabot), even though they should be cached. Example: https://github.com/apache/activemq-artemis/actions/runs/19174094055/job/54814185898 Yes, I'm aware that can happen to a small extent. The checks and test runs share caches, and the checks usually complete first, so it can be the case that certain things do still need downloaded as they didnt make it into the cache. But its pretty small compared to the overall dep load. I did used to segregate them, here and in other components jobs, but the disadvantage is that it doubles the cache usage, leading to more eviction churn for caches, especially across branches. They have increased the cache limits since then, but I still think its probably better as it is. Could probably also rig the 'checks' executions to trigger more inclusions with their 'extra time' buffer though. > > Well, at least this doesn't reach the 100k lines which happened after I changed the caching strategy :) Even though it's not extensive ... some bandwidth can be saved. > > About the T1C, I think i was looking at the wrong data before. But again as mentioned, it's the tests that take longer, and combined with the scrambled output I also don't think it's worth it. > > You mention certain tests needing "install" - I'll have to take a closer look at that, but I'm just bruteforcing through all the options and it seems the test have passed with "mvn verify" as well. Yes, because after a while of being annoyed by it, I did eventually change the build to _not actually fail_ if you didnt use "install" like it always had before. Check closer and youll find you shaved about a minute off the time, because it _didnt actually do exactly the same as it was before_. I'm signing out now, have a good weekend. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected] For further information, visit: https://activemq.apache.org/contact
