hadoop-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eli Collins <...@cloudera.com>
Subject Re: Release plans
Date Thu, 18 Feb 2010 19:28:31 GMT
On Thu, Feb 18, 2010 at 10:54 AM, Stack <stack@duboce.net> wrote:
> On Tue, Feb 16, 2010 at 2:12 PM, Eli Collins <eli@cloudera.com> wrote:
>> Hey guys,
>>
>> What are the current plans for the 21 release?
>>
> Doesn't seem like there is much interest going by the deafening silence Eli.
>
> Did it come up last night at the HUG?
>

It did, Owen gave an update in his talk and there was some follow on
discussion. The question of whether it makes sense to rebase 21 on
trunk came up, since 21 has gotten really out of date since it was
branched. What do you and others think of rebasing the 21 branch on
trunk once security is feature complete next month, and then targeting
an initial release a month after that? The rationale being that 21
branch is old, has not been getting the backports it needs, and the
community should be releasing relatively recent bits. In the mean time
we can focus on 21/trunk blockers. It doesn't look like there are a
ton more trunk blockers than 21 blockers. I know people, HBase in
particular, have been waiting for a 21 release for a while so delaying
further is a pain, and we should consider rolling what we've got, the
hope would be that the 21 branch would be more widely useful if it has
more recent features and fixes.

>> The roadmap on the wiki [1] says "minor releases are made regularly,
>> every few months." The 21 branch was created 5 months ago, and the
>> release dates in jira for the various core projects have all passed. I
>> know some projects, eg HBase, would like to get users on more recent
>> bits. What are the next steps for rolling a release?
>>
>
> Taking a look at blockers -- assuming to make a release, we just need
> to clear blockers and roll a candidate -- it seems like common/core
> and hdfs have but a few but mapreduce has a bunch.  Too many.
>
> Does anyone detect activity in the blocker issues?  I've not made a
> review.  Perhaps someone already has?

A decent number of the MR ones are patch available and quite a few
others are documentation related, nothing too scary. The core/hdfs
ones are pretty doable. We should make another pass to confirm that
they're all actually blockers. I'm happy to start working on the
blockers if that will help push the release forward.

Thanks,
Eli

Mime
View raw message