hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Enis Söztutar <e...@apache.org>
Subject Re: [VOTE] First release candidate for HBase 1.1.0 (RC0) is available.
Date Mon, 11 May 2015 03:55:19 GMT
Here are my tests so far:

checked sigs, crcs

build src tarsal with hadoop 2.3, 2.4.0, 2.5.0, 2.5.1, 2.5.2 and 2.6.0.
Build with Hadoop-2.2.0 is broken as previous mail. I don’t think we should
sink the RC for this.

run local mode

simple tests from shell

Build with downstreamer

checked dir layouts

checked jar files

checked version, tag,

Checked the documentation. Both the index.html and book are in the old
format. Nick did you copy the docs from master? This is unfortunate, but
not a blocker to the RC.

Run wiht LTT with 1M load on local mode and 5 node cluster. Seems fine.

Run with different block encoding and compression algorithms

Run 1.0.1 and 0.98.12 clients against 1.1.0 server, some smoke tests (list,
scan) and LTT

Going through the compat report:
https://people.apache.org/~enis/1.0.1_1.1.0RC0_compat_report.html, a couple
of interesting things:

  - https://issues.apache.org/jira/browse/HBASE-13551 missed 2 classes
related to proc v2. They should not be Public.

  - AuthUtil should not be Public.

This seems source incompatibility:

 - RegionScanner.nextRaw ( java.util.List<org.apache.hadoop.hbase.Cell> p1,
int p2 ) [abstract]  *:*  boolean

My vote would be -0, since the RegionScanner.nextRaw() although not used
much is a concerning change and breaks source compat.

Enis

On Wed, May 6, 2015 at 1:07 PM, Nick Dimiduk <ndimiduk@gmail.com> wrote:

> I'm also traveling today.
>
> I've already extended the vote for this RC to Sunday, and since no one has
> said this is a -1 -worthy regression, this candidate continues to stand.
>
> On Wed, May 6, 2015 at 12:16 PM, Andrew Purtell <andrew.purtell@gmail.com>
> wrote:
>
> > Formally, -0
> >
> > Given tomorrow is hbasecon perhaps it would be better to spin a RC on
> > Friday?
> >
> > I can take HBASE-13637 but am sitting on a plane at the moment. Won't be
> > able to get to it until tonight.
> >
> > > On May 6, 2015, at 10:43 AM, Nick Dimiduk <ndimiduk@apache.org> wrote:
> > >
> > > On Wed, May 6, 2015 at 10:13 AM, Andrew Purtell <
> > andrew.purtell@gmail.com>
> > > wrote:
> > >
> > >> I prefer to patch the POMs.
> > >
> > > Is this a formal -1?
> > >
> > > I've opened HBASE-13637 for tracking this issue. Let's get it fixed and
> > > I'll spin a new RC tonight.
> > >
> > >>> On May 5, 2015, at 4:16 PM, Nick Dimiduk <ndimiduk@gmail.com>
wrote:
> > >>>
> > >>> So what's the conclusion here? Are we dropping 2.2 support or
> updating
> > >> the
> > >>> poms and sinking the RC?
> > >>>
> > >>>> On Fri, May 1, 2015 at 7:47 AM, Sean Busbey <busbey@cloudera.com>
> > >> wrote:
> > >>>>
> > >>>> On Thu, Apr 30, 2015 at 6:48 PM, Andrew Purtell <
> apurtell@apache.org>
> > >>>> wrote:
> > >>>>
> > >>>>> We could patch our POMs to reference the hadoop-minikdc artifact
> > >>>>> independently of the rest of the Hadoop packages. It's standalone
> and
> > >>>>> rarely changes.
> > >>>> +1. I've been using HBase to test Hadoop changes for isolating
> > >> dependencies
> > >>>> from downstream folks (HADOOP-11804), and I've just been leaving
the
> > >>>> hadoop-minikdc artifact as-is due to these very reasons.
> > >>>>
> > >>>> --
> > >>>> Sean
> > >>
> >
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message