Closed briantopping closed 9 years ago
Thanks for the contributions! Nice to see the code being tidied up. I've made a couple of comments and once those are done i'll merge it and tag a release.
Thanks for the time to review this so quickly! It's always a pleasure to contribute when PRs don't go stale. :)
Please give me a bit to get the release plugin set up. I'll document how to use it and we can try a release from there, sound good?
Sounds great!
Can I go ahead and set up artifact publishing to Sonatype Maven repository? This isn't an official Apache project, so we can't release it there.
I wonder if the folk over at Mesosphere would be kind enough to host the project in their repos? They have a public maven repo over at http://downloads.mesosphere.io/maven. I'll ask around!
Oh wait, my bad:
<groupId>org.apache.mesos</groupId>
<artifactId>hadoop-mesos</artifactId>
<version>0.1.1-SNAPSHOT</version>
I can get this into the Apache repo if that's okay with the PMC.
Maybe send an email to the dev mailing list, or ping the IRC? #mesos
Pushing directly to Sonatype is the generally accepted clean path for small projects. If it's an Apache project, it should be pushed to the Apache repos. Both go directly to Maven Central.
If this isn't an Apache project, we should use a different GAV (group/artifact/version) combo.
WDYT?
I'm not sure if this is now technically an Apache project. It probably has that GAV because it was originally part of the Mesos source code and was brought out into it's own project by @brndnmtthws some time ago.
I think sending a message to the mailing list would be worth it, there's a bit of a wider discussion around this anyway related to hosting frameworks under /mesos
on github (and if we're continuing to do that, my vote goes for hosting it in the Apache repos).
Awesome. How would you feel about us taking this release configuration into a new PR and just keeping this one for the code cleanups/fixes?
Sounds like an excellent idea. I've rolled back the version change per your request.
In my deployment of 2.6.0-mr1-cdh5.4.0, I had a problem where "2015-05-08 03:33:24,421 INFO org.apache.hadoop.mapred.MesosScheduler: Unknown/exited TaskTracker: http://10.211.55.16:50060. " was being emitted 3-5 times a second. I don't have a JWDP connection figured out to see what the messages were, but I'm guessing there is an internal housekeeping job that is used internally. This implementation allows that to happen quietly.