Fwd: [DISCUSS] EOL Hadoop branch-2.8

[Wei-Chiu Chuang <we...@cloudera.com.INVALID>]

FYI
Also filed HBASE-23893 <https://issues.apache.org/jira/browse/HBASE-23893> to
keep track of the relevant work in HBase.

I'd like to gather the feedback from the HBase community. Once branch-2.8
is EOL, what's the plan of Hadoop 2.x support?
IMHO Hadoop 2.9 EOL would be the next. The last 2.9 release was 15 months
ago and doesn't look like any one's picking it up.

I don't think HBase is certified to run on Hadoop 2.10 yet.
But in my opinion, I'd like to see Hadoop reducing Hadoop 2.x footprint
sooner than later.

---------- Forwarded message ---------
From: Wei-Chiu Chuang <we...@apache.org>
Date: Mon, Feb 17, 2020 at 9:44 AM
Subject: [DISCUSS] EOL Hadoop branch-2.8
To: Hadoop Common <co...@hadoop.apache.org>, Hdfs-dev <
hdfs-dev@hadoop.apache.org>, yarn-dev <ya...@hadoop.apache.org>,
mapreduce-dev <ma...@hadoop.apache.org>


The last Hadoop 2.8.x release, 2.8.5, was GA on September 15th 2018.

It's been 17 months since the release and the community by and large have
moved up to 2.9/2.10/3.x.

With Hadoop 3.3.0 over the horizon, is it time to start the EOL discussion
and reduce the number of active branches?

Re: [DISCUSS] EOL Hadoop branch-2.8

[Sean Busbey <bu...@apache.org>]

Historically our practice has been that once we determine a Hadoop release
line is EOM[1], the next minor release of HBase moves that release line to
"you should avoid this" status[2] and we stop any automated testing against
it.

I could be mistaken, but I think our habits around new minor Hadoop 2
releases of late have been pretty minimal testing. Essentially we check
compilation and do a simple integration check with HDFS.

Sometimes a community member comes forward with some more in depth testing
like ITBLL runs or notes about the Hadoop they use in production
personally, but afaik we don't aggregate that information anywhere.

We haven't really hashed out as a community what we'll do if Hadoop stops
making Hadoop 2 releases entirely. I think we're implicitly all hoping a
single maintenance line continues until we can mark all HBase 1.y EOM.


[1] when there's an official announcement from Hadoop about EOM a release
line we try to note that specifically. When Hadoop fails to address a
public CVE in a release line then things are a bit messier


[2] the section in the ref guide is "Hadoop" under "basic prerequisites"

http://hbase.apache.org/book.html#hadoop

There's a set of notes below the table that tries to give some reasoning
for the "avoid" status.


On Tue, Feb 25, 2020, 16:44 Wei-Chiu Chuang <we...@cloudera.com.invalid>
wrote:

> FYI
> Also filed HBASE-23893 <https://issues.apache.org/jira/browse/HBASE-23893>
> to
> keep track of the relevant work in HBase.
>
> I'd like to gather the feedback from the HBase community. Once branch-2.8
> is EOL, what's the plan of Hadoop 2.x support?
> IMHO Hadoop 2.9 EOL would be the next. The last 2.9 release was 15 months
> ago and doesn't look like any one's picking it up.
>
> I don't think HBase is certified to run on Hadoop 2.10 yet.
> But in my opinion, I'd like to see Hadoop reducing Hadoop 2.x footprint
> sooner than later.
>
> ---------- Forwarded message ---------
> From: Wei-Chiu Chuang <we...@apache.org>
> Date: Mon, Feb 17, 2020 at 9:44 AM
> Subject: [DISCUSS] EOL Hadoop branch-2.8
> To: Hadoop Common <co...@hadoop.apache.org>, Hdfs-dev <
> hdfs-dev@hadoop.apache.org>, yarn-dev <ya...@hadoop.apache.org>,
> mapreduce-dev <ma...@hadoop.apache.org>
>
>
> The last Hadoop 2.8.x release, 2.8.5, was GA on September 15th 2018.
>
> It's been 17 months since the release and the community by and large have
> moved up to 2.9/2.10/3.x.
>
> With Hadoop 3.3.0 over the horizon, is it time to start the EOL discussion
> and reduce the number of active branches?
>