accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Josh Elser (JIRA)" <>
Subject [jira] [Commented] (ACCUMULO-1797) update release practices to cover multiple Hadoops
Date Tue, 22 Oct 2013 00:10:41 GMT


Josh Elser commented on ACCUMULO-1797:

I'm not sure I agree with what you said [~kturner]. (When I say "we" going forward, I mean
the collective Accumulo community, which, when it comes down to it, likely refers to committers.)
We haven't really had a good track record for involvement in release testing. Perhaps making
a more discrete testing plan for committers to follow on their "flavors of choice" (OS and
Java, Hadoop, Zookeeper versions) would get committers more involved?

First off, I think that we should be testing Accumulo against each "stable" version of Hadoop
at the time of that release -- so, for 1.6.0, this equates to 1.2.1 and 2.2.0. This also defines
what I would call "supported". Unless the Hadoop community would do something drastic that
would break us (which I would never expect to be the case), I think tracking Apache Hadoop
would be the best. I think this also makes management of the hadoop.profile's a bit easier
to track/maintain.

Zookeeper, I'm less concerned about testing. I'd feel ok testing against the most recent,
stable release (3.4.5, presently), but if someone wants to test 3.3.6 too, that'd be good.
I just wouldn't want to hold up our releases for not having rigorous testing against 3.3.6.

> update release practices to cover multiple Hadoops
> --------------------------------------------------
>                 Key: ACCUMULO-1797
>                 URL:
>             Project: Accumulo
>          Issue Type: Sub-task
>            Reporter: Sean Busbey
> if we're going to label Hadoop platforms as supported, we should ensure the release process
specifies that tests are run against each of them.

This message was sent by Atlassian JIRA

View raw message