I took a look at items in 2.3 and 2.4, as well as CDH5 and HDP2 (also
looked at a few of the patches to assess their risk levels), and came
up with the following strawman propose of bug-patches to be included
in a 2.2.1 release:
HADOOP-10029 [major] - Specifying har file to MR job fails in secure cluster
HDFS-5089 [major] - When a LayoutVersion supports SNAPSHOT, it must
support FSIMAGE_NAME_OPTIMIZATION
HDFS-5403 [major] - WebHdfs client cannot communicate with older
WebHdfs servers post HDFS-5306
HDFS-5433 [critical] - When reloading fsimage during checkpointing, we
should clear existing snapshottable directories
MAPREDUCE-5028 [critical] - Maps fail when io.sort.mb is set to high value
YARN-1295 [major] - In UnixLocalWrapperScriptBuilder, using bash -c
can cause Text file busy errors
YARN-1374 [blocker] - Resource Manager fails to start due to
ConcurrentModificationException
YARN-1176 [critical] - RM web services ClusterMetricsInfo total nodes
doesn't include unhealthy nodes
There are lots of outstanding bug fixes, so this list is definitely a
bit arbitrary, but it seemed like a good list to me. Any thoughts?
On Fri, Jan 3, 2014 at 5:26 PM, Sandy Ryza <sandy.r...@cloudera.com> wrote:
> Re-reading the thread, it seems what I said about 2.2.1 never happening was
> incorrect. My impression is still that nobody has plans to drive a 2.2.1
> release on any particular timeline.
>
> The changes that are now in 2.3 have been moved out of the branch-2.2.1. I
> suppose the idea is that changes slated for 2.2.1 should be committed both
> to branch-2.2 and branch-2.2.1.
>
> -Sandy
>
>
> On Fri, Jan 3, 2014 at 4:57 PM, Raymie Stata <rst...@altiscale.com> wrote:
>
>> Yes, that thread is part of what's confusing me. Arun's initial 11/8
>> message suggests that there would be room for blocker fixes leading to
>> a 2.2.1 patch release ("...and then be very careful about including
>> only *blocker* fixes in branch-2.2"). And nothing else in that thread
>> suggests that there wouldn't be a patch release. And yet, Sandy seems
>> to think that "2.2.1 isn't happening at all" (YARN-1295), a view
>> that's consistent with the currently confused state of the repo
>> (branch-2.2.1 exists but not released, branch-2.2 version is
>> 2.2.2-SNAPSHOT).
>>
>> Seems to me that we should be planning for a 2.2.1 patch release at
>> some point...
>>
>> Raymie
>>
>>
>>
>> On Fri, Jan 3, 2014 at 1:17 AM, Steve Loughran <ste...@hortonworks.com>
>> wrote:
>> > the last discussion on this was in november -I presume that's still the
>> plan
>> >
>> >
>> http://mail-archives.apache.org/mod_mbox/hadoop-common-dev/201311.mbox/%3CA31E1430-33BE-437C-A61E-050F9A67C109%40hortonworks.com%3E
>> >
>> >
>> > On 3 January 2014 04:10, Raymie Stata <rst...@altiscale.com> wrote:
>> >
>> >> Nudge, any thoughts?
>> >>
>> >> On Sun, Dec 29, 2013 at 1:25 AM, Raymie Stata <rst...@altiscale.com>
>> >> wrote:
>> >> > In discussing YARN-1295 it's become clear that I'm confused about the
>> >> > outcome of the "Next releases" thread. I had assumed there would be
>> >> > patch releases to 2.2, and indeed one would be coming out early Q1.
>> >> > Is this correct?
>> >> >
>> >> > If so, then things seem a little messed-up right now in 2.2-land.
>> >> > There already is a branch-2.2.1, but there hasn't been a release. And
>> >> > branch-2.2 has Maven version 2.2.2-SNAPSHOT. Due to the "2.3 rename"
>> >> > a few weeks ago, it might be that the first patch release for 2.2
>> >> > needs to be 2.2.2. But if so, notice these lists of fixes for 2.2.1:
>> >> >
>> >> > https://issues.apache.org/jira/browse/YARN/fixforversion/12325667
>> >> > https://issues.apache.org/jira/browse/HDFS/fixforversion/12325666
>> >> >
>> >> > Do these need to have their fix-versions updated?
>> >> >
>> >> > Raymie
>> >> >
>> >> >
>> >> > P.S. While we're on the subject of point releases, let me check my
>> >> assumptions.
>> >> >
>> >> > I assumed that, for release x.y.z, fixes deemed to be "critical bug
>> >> > fixes" would be put into branch-x.y as a matter of course. The Maven
>> >> > release-number in branch-x.y would be x.y.(z+1)-SNAPSHOT, and JIRAs
>> >> > (to be) committed to branch-x.y would have x.y.(z+1) as one of their
>> >> > fix-versions.
>> >> >
>> >> > When enough fixes have accumulated to warrant a release, or when a fix
>> >> > comes up that is critical enough to warrant an immediate release, then
>> >> > branch-x-y is branched to branch-x.y.(z+1), and a release is made.
>> >> >
>> >> > (As Hadoop itself moves from x.y to x.(y+1) and then x.(y+2), the
>> >> > threshold for what is considered to be a "critical bug" would
>> >> > naturally start to rise, as the effort of back-porting goes up.)
>> >> >
>> >> > Do I have it right?
>> >>
>> >
>> > --
>> > CONFIDENTIALITY NOTICE
>> > NOTICE: This message is intended for the use of the individual or entity
>> to
>> > which it is addressed and may contain information that is confidential,
>> > privileged and exempt from disclosure under applicable law. If the reader
>> > of this message is not the intended recipient, you are hereby notified
>> that
>> > any printing, copying, dissemination, distribution, disclosure or
>> > forwarding of this communication is strictly prohibited. If you have
>> > received this communication in error, please contact the sender
>> immediately
>> > and delete it from your system. Thank You.
>>
