Hi Chris,
Thanks for your response!
I think I could miss the thread discussion of "[DISCUSS] 2.6.x line
releases" for something reason. I checked the discussion - Sean claimed that
HBase community needs 2.6.5, Zhe said they are using 2.6.x releases and Akira
said that are over new 50 commits land on branch-2.6 since 2.6.4. Do I miss any
comments there?
These comments are more like wishes but not giving more clarifications on
the needs. I would like to hear more specific reasons to not move to 2.7.x
releases but prefer to upgrade to 2.6.5. If the only reason is about
expectation management, I think we should claim 2.6.5 is the last branch-2.6
release after this release work, otherwise people would expect us to maintain
this branch forever which is impossible and unnecessary. Thoughts?
Thanks,
Junping
________________________________
From: Chris Trezzo <ctre...@gmail.com>
Sent: Wednesday, August 10, 2016 9:30 PM
To: Junping Du
Cc: common-...@hadoop.apache.org; hdfs-...@hadoop.apache.org;
mapreduce-dev@hadoop.apache.org; yarn-...@hadoop.apache.org; Jason Lowe
Subject: Re: [Release thread] 2.6.5 release activities
Thanks Jason and Junping for the comments! I will update the spreadsheet for
HADOOP-13362 and YARN-4794.
As for continuing 2.6.x releases, please see the discussion in the "[DISCUSS]
2.6.x line releases" thread. Sean, Akira and Zhe all expressed interest in
additional 2.6.x releases. I started this thread based off of that interest. I
understand there is a burden to maintaining a large number of branches. I am
not sure what the community's end-of-life policy is, but maybe we can issue a
warning with the 2.6.5 release stating when we will stop maintaining the
release line. This at least gives users some time to make migration plans to a
newer version.
On Wed, Aug 10, 2016 at 9:36 AM, Junping Du
<j...@hortonworks.com<mailto:j...@hortonworks.com>> wrote:
Thanks Chris for bring up this discussion.
Before we going to detail discussion of releasing 2.6.5, I have a quick
question here: do we think it is necessary to continue to release branch-2.6,
like 2.6.5, etc after 2.7 is out for more than 1 year. Any reason to not
suggest users to upgrade to 2.7.3 releases for latest fixes which is in
releasing now?
My major concern on more release efforts on legacy branches is the same with my
comments on other release plan before - it seems too many releases trains get
planned at the same time window (2.6.x, 2.7.x, 2.8, 3.0-alpha, 3.1-beta, etc.).
Not only user could get confusing on this, but also I suspect we don't have so
many bandwidth in community to push forward so these releases in high quality
during the same time window - just like Chris Douglas mentioned in another
email thread on committer activity and bandwidth. IMO, may be it is better to
focus on limited number of releases and move them faster?
BTW, I agree with Jason that HADOOP-13362 is not needed for branch-2.6 unless
we backport container metrics related patches there.
Thanks,
Junping
________________________________________
From: Jason Lowe <jl...@yahoo-inc.com.INVALID>
Sent: Wednesday, August 10, 2016 4:14 PM
To: Chris Trezzo;
common-...@hadoop.apache.org<mailto:common-...@hadoop.apache.org>;
hdfs-...@hadoop.apache.org<mailto:hdfs-...@hadoop.apache.org>;
mapreduce-dev@hadoop.apache.org<mailto:mapreduce-dev@hadoop.apache.org>;
yarn-...@hadoop.apache.org<mailto:yarn-...@hadoop.apache.org>
Subject: Re: [Release thread] 2.6.5 release activities
Thanks for organizing this, Chris!
I don't believe HADOOP-13362 is needed since it's related to ContainerMetrics.
ContainerMetrics weren't added until 2.7 by YARN-2984.
YARN-4794 looks applicable to 2.6. The change drops right in except it has
JDK7-isms (multi-catch clause), so it needs a slight change.
Jason
From: Chris Trezzo <ctre...@gmail.com<mailto:ctre...@gmail.com>>
To: "common-...@hadoop.apache.org<mailto:common-...@hadoop.apache.org>"
<common-...@hadoop.apache.org<mailto:common-...@hadoop.apache.org>>;
hdfs-...@hadoop.apache.org<mailto:hdfs-...@hadoop.apache.org>;
"mapreduce-dev@hadoop.apache.org<mailto:mapreduce-dev@hadoop.apache.org>"
<mapreduce-dev@hadoop.apache.org<mailto:mapreduce-dev@hadoop.apache.org>>;
"yarn-...@hadoop.apache.org<mailto:yarn-...@hadoop.apache.org>"
<yarn-...@hadoop.apache.org<mailto:yarn-...@hadoop.apache.org>>
Sent: Tuesday, August 9, 2016 7:32 PM
Subject: [Release thread] 2.6.5 release activities
Based on the sentiment in the "[DISCUSS] 2.6.x line releases" thread, I
have moved forward with some of the initial effort in creating a 2.6.5
release. I am forking this thread so we have a dedicated 2.6.5 release
thread.
I have gone through the git logs and gathered a list of JIRAs that are in
branch-2.7 but are missing from branch-2.6. I limited the diff to issues
with a commit date after 1/26/2016. I did this because 2.6.4 was cut from
branch-2.6 around that date (http://markmail.org/message/xmy7ebs6l3643o5e)
and presumably issues that were committed to branch-2.7 before then were
already looked at as part of 2.6.4.
I have collected these issues in a spreadsheet and have given them an
initial triage on whether they are candidates for a backport to 2.6.5. The
spreadsheet is sorted by the status of the issues with the potential
backport candidates at the top. Here is a link to the spreadsheet:
https://docs.google.com/spreadsheets/d/1lfG2CYQ7W4q3olWpOCo6EBAey1WYC8hTRUemHvYPPzY/edit?usp=sharing
As of now, I have identified 16 potential backport candidates. Please take
a look at the list and let me know if there are any that you think should
not be on the list, or ones that you think I have missed. This was just an
initial high-level triage, so there could definitely be issues that are
miss-labeled.
As a side note: we still need to look at the pre-commit build for 2.6 and
follow up with an addendum for HADOOP-12800.
Thanks everyone!
Chris Trezzo
