You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-dev@hadoop.apache.org by Matt Foley <ma...@apache.org> on 2011/12/08 20:57:49 UTC
[VOTE] Hadoop-1.0.0 release candidate 2
Hello all,
I have posted a release candidate for Hadoop 1.0.0 at
http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
Please download, evaluate, and vote on this list.
The artifacts will be posted to the maven repo shortly.
The vote will close at noon PST on Thursday 15 Dec.
It may be of interest that it is important to use an appropriate version of
Java.
I lost several days due to instabilities apparently in Oracle JDK 1.6.0_23,
which
caused about 30 junit test failures in contrib (streaming, schedulers, and
gridmix).
Switching to JDK 1.6.0_26 made the problems go away.
The release artifacts in this hadoop-1.0.0-rc2 were build with JDK 1.6.0_26.
Thank you,
--Matt (Release Manager for 1.0.0, formerly known as 0.20.205.1)
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Fri, Dec 9, 2011 at 6:29 PM, Ashutosh Chauhan <ha...@apache.org> wrote:
> Hey Roman,
>
> That problem is at Hive end. You can try out the patch at
> https://issues.apache.org/jira/browse/HIVE-2631 to unblock your testing.
Sure. As I said -- the fix is pretty simple. However, releasing this
would break every single existing Hive installation out there.
Is this a good thing? Should we message it somehow? Should
we coordinate with Hive folks?
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
Yes, I agree that we, the whole Hadoop development community, need to help
each other get the various components working together, and keep them
mutually compatible as each continues to grow and evolve. I think BigTop
is providing a valuable service in this area, by providing an official
Apache place for the components to be tested together.
--Matt
On Mon, Dec 12, 2011 at 8:20 PM, Roman Shaposhnik <rv...@apache.org> wrote:
> On Sat, Dec 10, 2011 at 12:24 PM, Matt Foley <mf...@hortonworks.com>
> wrote:
> >>> Btw, I think it is very instructive to take a look at how
> > Linux community has dealt
> >>> with version transition between 2.X -> 3.X.
> >
> > You mean you want us to not call it 1.0.0? As I remember the vote was
> > overwhelmingly
> > in favor of calling it 1.0.0.
>
> That's not what I was trying to drive at. Changing versions is fine
> as long as there's a clear messaging around it to downstream.
>
> For example, making sure that HIVE-2631 is scheduled for
> inclusion into an upcoming Hive 0.8.0 release would greatly
> those types of concerns for Hive.
>
> Also having a clear message on a website/etc would help.
>
> On the other hand -- NOT changing versions for quite
> a bit of time (to let downstream catch up) is also an
> option. That's what experience with Linux versioning
> has taught me.
>
> Thanks,
> Roman.
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Sat, Dec 10, 2011 at 12:24 PM, Matt Foley <mf...@hortonworks.com> wrote:
>>> Btw, I think it is very instructive to take a look at how
> Linux community has dealt
>>> with version transition between 2.X -> 3.X.
>
> You mean you want us to not call it 1.0.0? As I remember the vote was
> overwhelmingly
> in favor of calling it 1.0.0.
That's not what I was trying to drive at. Changing versions is fine
as long as there's a clear messaging around it to downstream.
For example, making sure that HIVE-2631 is scheduled for
inclusion into an upcoming Hive 0.8.0 release would greatly
those types of concerns for Hive.
Also having a clear message on a website/etc would help.
On the other hand -- NOT changing versions for quite
a bit of time (to let downstream catch up) is also an
option. That's what experience with Linux versioning
has taught me.
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
>> Btw, I think it is very instructive to take a look at how
Linux community has dealt
>> with version transition between 2.X -> 3.X.
You mean you want us to not call it 1.0.0? As I remember the vote was
overwhelmingly
in favor of calling it 1.0.0.
--Matt
On Sat, Dec 10, 2011 at 11:23 AM, Roman Shaposhnik <rv...@apache.org> wrote:
> On Fri, Dec 9, 2011 at 6:29 PM, Ashutosh Chauhan <ha...@apache.org>
> wrote:
> > Hey Roman,
> >
> > That problem is at Hive end. You can try out the patch at
> > https://issues.apache.org/jira/browse/HIVE-2631 to unblock your testing.
>
> Btw, I think it is very instructive to take a look at how Linux
> community has dealt
> with version transition between 2.X -> 3.X. Here's a thread that I still
> have in
> my short term memory. There were, of course, others
>
> https://plus.google.com/106327083461132854143/posts/SbnL3KaVRtM#106327083461132854143/posts/SbnL3KaVRtM
>
> Thanks,
> Roman.
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Fri, Dec 9, 2011 at 6:29 PM, Ashutosh Chauhan <ha...@apache.org> wrote:
> Hey Roman,
>
> That problem is at Hive end. You can try out the patch at
> https://issues.apache.org/jira/browse/HIVE-2631 to unblock your testing.
Btw, I think it is very instructive to take a look at how Linux
community has dealt
with version transition between 2.X -> 3.X. Here's a thread that I still have in
my short term memory. There were, of course, others
https://plus.google.com/106327083461132854143/posts/SbnL3KaVRtM#106327083461132854143/posts/SbnL3KaVRtM
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Ashutosh Chauhan <ha...@apache.org>.
Hey Roman,
That problem is at Hive end. You can try out the patch at
https://issues.apache.org/jira/browse/HIVE-2631 to unblock your testing.
Thanks,
Ashutosh
On Fri, Dec 9, 2011 at 18:21, Roman Shaposhnik <rv...@apache.org> wrote:
> +Carl (the only person from Hive project I know ;-))
>
> On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> > Hello all,
> > I have posted a release candidate for Hadoop 1.0.0 at
> > http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>
> Ok, my test run is now past HBase (horray! that works).
>
> That's the good news so far. The bad news is -- this release
> is incompatible with hive:
>
> $ hive
> Hive requires Hadoop 0.20.x (x >= 1).
> 'hadoop version' returned:
> Hadoop 1.0.0.2 Subversion -r Compiled by jenkins on Fri Dec 9 13:29:12 EST
> 2011
>
> This is, of course, a minor, but a VERY annoying thing.
>
> Thoughts on what to do?
>
> Thanks,
> Roman.
>
> P.S. I wish the test run took less than half a day :-(
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
+Carl (the only person from Hive project I know ;-))
On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> Hello all,
> I have posted a release candidate for Hadoop 1.0.0 at
> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
Ok, my test run is now past HBase (horray! that works).
That's the good news so far. The bad news is -- this release
is incompatible with hive:
$ hive
Hive requires Hadoop 0.20.x (x >= 1).
'hadoop version' returned:
Hadoop 1.0.0.2 Subversion -r Compiled by jenkins on Fri Dec 9 13:29:12 EST 2011
This is, of course, a minor, but a VERY annoying thing.
Thoughts on what to do?
Thanks,
Roman.
P.S. I wish the test run took less than half a day :-(
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Fri, Dec 9, 2011 at 4:30 PM, Matt Foley <mf...@hortonworks.com> wrote:
> Roman, the 1.0.0-rc2 is built from branch-1.0@1211738, which contains
> c1207581.
Thanks. This is useful to know. Wish I had that info earlier :-(
> Is the problem consistently reproducible in 1.0.0-rc2? Perhaps you're
> seeing some intermittent behavior?
Well, when I tried to reproduce it (and couldn't) I was running a
pseudo-distributed
setup on my latop. When I see the problem -- it is on a fully
distributed cluster
running off of Bigtop Jenkins:
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-rc-smoketest/
This *may* have something to do with it.
Anyway, now that we have the patch that actually fixes the issue 100%, the
real question is -- whether we should apply it to RC.
Thanks,
Roman.
P.S. In the future I'll try to reproduce everything over @Bigtop jenkins so that
at least we have a "paper trail" of things. Running tests over there is
pretty cheap.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
Roman, the 1.0.0-rc2 is built from branch-1.0@1211738, which contains
c1207581.
Is the problem consistently reproducible in 1.0.0-rc2? Perhaps you're
seeing
some intermittent behavior?
On Fri, Dec 9, 2011 at 3:57 PM, Roman Shaposhnik <rv...@apache.org> wrote:
> On Fri, Dec 9, 2011 at 3:26 PM, Matt Foley <mf...@hortonworks.com> wrote:
> > Hi Roman,
> > regarding MAPREDUCE-3319 I'm confused too. Your comment in the bug as
> > of 28/Nov/11 21:49 seems to indicate that it is fixed, but clearly it
> wasn't
> > fixed the same way as Subroto's patch (which came in on Dec 6).
>
> What is the tag corresponding to this RC? The svn rev. where I couldn't
> reproduce it is this: branch-1.0@1207581 not sure how it relates to the
> rev
> from which the RC was cut. That's why I asked this question on a JIRA:
>
> https://issues.apache.org/jira/browse/MAPREDUCE-3319?focusedCommentId=13158818&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13158818
>
> > Anyway, I reviewed and committed Subroto's patch to branch-1. If there
> is a
> > need to regenerate the RC for 1.0.0, I'll merge it to branch-1.0, but I
> hope
> > no one will think this is serious enough to be a cause to regenerate the
> RC
> > just for this patch.
>
> It would be sad to see 1.0 with a bundled example that doesn't work out of
> the box because of a trivial problem. There's only been one day of RC
> out there and the fix seems very low risk. Given that, perhaps the
> trade-off of wasting one day of potential testing of RC by a community
> can be considered.
>
> I can offer any help with regenerating the RC with patch applied that you
> might need.
>
> Thanks,
> Roman.
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Fri, Dec 9, 2011 at 3:26 PM, Matt Foley <mf...@hortonworks.com> wrote:
> Hi Roman,
> regarding MAPREDUCE-3319 I'm confused too. Your comment in the bug as
> of 28/Nov/11 21:49 seems to indicate that it is fixed, but clearly it wasn't
> fixed the same way as Subroto's patch (which came in on Dec 6).
What is the tag corresponding to this RC? The svn rev. where I couldn't
reproduce it is this: branch-1.0@1207581 not sure how it relates to the rev
from which the RC was cut. That's why I asked this question on a JIRA:
https://issues.apache.org/jira/browse/MAPREDUCE-3319?focusedCommentId=13158818&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13158818
> Anyway, I reviewed and committed Subroto's patch to branch-1. If there is a
> need to regenerate the RC for 1.0.0, I'll merge it to branch-1.0, but I hope
> no one will think this is serious enough to be a cause to regenerate the RC
> just for this patch.
It would be sad to see 1.0 with a bundled example that doesn't work out of
the box because of a trivial problem. There's only been one day of RC
out there and the fix seems very low risk. Given that, perhaps the
trade-off of wasting one day of potential testing of RC by a community
can be considered.
I can offer any help with regenerating the RC with patch applied that you
might need.
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Marco Zühlke <mz...@gmail.com>.
Chris K Wensel <ch...@...> writes:
>
> fwiw, looks like there is a dependency missing for jackson in the published
pom file.
Hi Chris,
there is already an issues for this: HADOOP-7461
Marco
>
> Looks like JobClient imports some of its classes causing Hadoop local mode to
fail if running directly from
> a mvn retrieve etc..
>
> There seems to be a fair number of Cascading/Hadoop sub-projects that do this
a lot.
>
> chris
>
> --
> Chris K Wensel
> chris@...
> http://concurrentinc.com
>
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Chris K Wensel <ch...@wensel.net>.
fwiw, looks like there is a dependency missing for jackson in the published pom file.
Looks like JobClient imports some of its classes causing Hadoop local mode to fail if running directly from a mvn retrieve etc..
There seems to be a fair number of Cascading/Hadoop sub-projects that do this a lot.
chris
On Dec 9, 2011, at 3:26 PM, Matt Foley wrote:
> Hi Roman,
> regarding MAPREDUCE-3319<https://issues.apache.org/jira/browse/MAPREDUCE-3319>
> I'm
> confused too. Your comment in the bug as of 28/Nov/11 21:49 seems to
> indicate that it is fixed, but clearly it wasn't fixed the same way as
> Subroto's patch (which came in on Dec 6).
>
> Anyway, I reviewed and committed Subroto's patch to branch-1. If there is
> a need to regenerate the RC for 1.0.0, I'll merge it to branch-1.0, but I
> hope no one will think this is serious enough to be a cause to regenerate
> the RC just for this patch.
>
> BTW, the same change would be applicable to trunk, so I've left the bug
> open.
>
> --Matt
>
> On Fri, Dec 9, 2011 at 2:47 PM, Roman Shaposhnik <rv...@apache.org> wrote:
>
>> On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
>>> Hello all,
>>> I have posted a release candidate for Hadoop 1.0.0 at
>>> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>>>
>>> Please download, evaluate, and vote on this list.
>>
>> Pulled into Bigtop, built and running integrations tests today.
>>
>> Quick question though: how come this one is not fixed?
>> https://issues.apache.org/jira/browse/MAPREDUCE-3319
>> There's a patch posted, etc *and* it wasn't reproducible
>> in the branch. I'm confused.
>>
>> Thanks,
>> Roman.
>>
--
Chris K Wensel
chris@concurrentinc.com
http://concurrentinc.com
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
Hi Roman,
regarding MAPREDUCE-3319<https://issues.apache.org/jira/browse/MAPREDUCE-3319>
I'm
confused too. Your comment in the bug as of 28/Nov/11 21:49 seems to
indicate that it is fixed, but clearly it wasn't fixed the same way as
Subroto's patch (which came in on Dec 6).
Anyway, I reviewed and committed Subroto's patch to branch-1. If there is
a need to regenerate the RC for 1.0.0, I'll merge it to branch-1.0, but I
hope no one will think this is serious enough to be a cause to regenerate
the RC just for this patch.
BTW, the same change would be applicable to trunk, so I've left the bug
open.
--Matt
On Fri, Dec 9, 2011 at 2:47 PM, Roman Shaposhnik <rv...@apache.org> wrote:
> On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> > Hello all,
> > I have posted a release candidate for Hadoop 1.0.0 at
> > http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
> >
> > Please download, evaluate, and vote on this list.
>
> Pulled into Bigtop, built and running integrations tests today.
>
> Quick question though: how come this one is not fixed?
> https://issues.apache.org/jira/browse/MAPREDUCE-3319
> There's a patch posted, etc *and* it wasn't reproducible
> in the branch. I'm confused.
>
> Thanks,
> Roman.
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> Hello all,
> I have posted a release candidate for Hadoop 1.0.0 at
> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>
> Please download, evaluate, and vote on this list.
Pulled into Bigtop, built and running integrations tests today.
Quick question though: how come this one is not fixed?
https://issues.apache.org/jira/browse/MAPREDUCE-3319
There's a patch posted, etc *and* it wasn't reproducible
in the branch. I'm confused.
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Andrew Purtell <ap...@apache.org>.
See the last comment on HADOOP-7929.
My working style is to put up a patch that sketches what changes I am thinking, if this is not too labor intensive to produce, even if while concurrently running tests to confirm assumptions. If I have something real, I'll set 'Patch Available'. Apologies if this caused any confusion here, and I can do this differently going forward. (?)
I _still_ don't have a working secure-everything environment with the latest RCs of all of the pieces, but 7929 is a red herring.
Best regards,
- Andy
Problems worthy of attack prove their worth by hitting back. - Piet Hein (via Tom White)
----- Original Message -----
> From: Stack <st...@duboce.net>
> To: common-dev@hadoop.apache.org
> Cc:
> Sent: Friday, December 16, 2011 9:56 AM
> Subject: Re: [VOTE] Hadoop-1.0.0 release candidate 2
>
> On Fri, Dec 16, 2011 at 12:32 AM, Matt Foley <mf...@hortonworks.com>
> wrote:
>> Dear all,
>> I have a clean build of the below, that passes junit testing, and was about
>> to post it, when I got email about the newly opened
>> HADOOP-7929<https://issues.apache.org/jira/browse/HADOOP-7929>
> "Port
>> HADOOP-7070 to branch-1". It appears this prevents secure HBase from
>> working with secure Hadoop. Since HBase support is a key element of 1.0.0,
>> and it seems this will be fixable fairly promptly, I am going to wait
>> another day for 1.0.0-rc3.
>>
>
> Thanks Matt. Yes, secure hbase needs this.
> St.Ack
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Stack <st...@duboce.net>.
On Fri, Dec 16, 2011 at 12:32 AM, Matt Foley <mf...@hortonworks.com> wrote:
> Dear all,
> I have a clean build of the below, that passes junit testing, and was about
> to post it, when I got email about the newly opened
> HADOOP-7929<https://issues.apache.org/jira/browse/HADOOP-7929> "Port
> HADOOP-7070 to branch-1". It appears this prevents secure HBase from
> working with secure Hadoop. Since HBase support is a key element of 1.0.0,
> and it seems this will be fixable fairly promptly, I am going to wait
> another day for 1.0.0-rc3.
>
Thanks Matt. Yes, secure hbase needs this.
St.Ack
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
Dear all,
I have a clean build of the below, that passes junit testing, and was about
to post it, when I got email about the newly opened
HADOOP-7929<https://issues.apache.org/jira/browse/HADOOP-7929> "Port
HADOOP-7070 to branch-1". It appears this prevents secure HBase from
working with secure Hadoop. Since HBase support is a key element of 1.0.0,
and it seems this will be fixable fairly promptly, I am going to wait
another day for 1.0.0-rc3.
Thanks for your patience,
--Matt
On Tue, Dec 13, 2011 at 6:17 PM, Matt Foley <mf...@hortonworks.com> wrote:
> In various emails and jiras, I have been asked to incorporate the
> following additional patches in 1.0.0:
>
> - MAPREDUCE-3319 (Roman)
> - HADOOP-7903 (Arpit)
> - MAPREDUCE-3475 (Daryn)
> - HDFS-2589 (Daryn)
> - HADOOP jira about to be opened (Chris W.) for missing jackson
> dependency in hadoop pom.
>
> I have not heard of any other significant issues in this build.
> I will re-spin the release candidate tomorrow.
>
> Thanks,
> --Matt
>
> On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
>
>> Hello all,
>> I have posted a release candidate for Hadoop 1.0.0 at
>> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>>
>> Please download, evaluate, and vote on this list.
>> The artifacts will be posted to the maven repo shortly.
>> The vote will close at noon PST on Thursday 15 Dec.
>>
>> It may be of interest that it is important to use an appropriate version
>> of Java.
>> I lost several days due to instabilities apparently in Oracle JDK
>> 1.6.0_23, which
>> caused about 30 junit test failures in contrib (streaming, schedulers,
>> and gridmix).
>> Switching to JDK 1.6.0_26 made the problems go away.
>>
>> The release artifacts in this hadoop-1.0.0-rc2 were build with
>> JDK 1.6.0_26.
>>
>> Thank you,
>> --Matt (Release Manager for 1.0.0, formerly known as 0.20.205.1)
>>
>
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Matt Foley <mf...@hortonworks.com>.
In various emails and jiras, I have been asked to incorporate the following
additional patches in 1.0.0:
- MAPREDUCE-3319 (Roman)
- HADOOP-7903 (Arpit)
- MAPREDUCE-3475 (Daryn)
- HDFS-2589 (Daryn)
- HADOOP jira about to be opened (Chris W.) for missing jackson
dependency in hadoop pom.
I have not heard of any other significant issues in this build.
I will re-spin the release candidate tomorrow.
Thanks,
--Matt
On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> Hello all,
> I have posted a release candidate for Hadoop 1.0.0 at
> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>
> Please download, evaluate, and vote on this list.
> The artifacts will be posted to the maven repo shortly.
> The vote will close at noon PST on Thursday 15 Dec.
>
> It may be of interest that it is important to use an appropriate version
> of Java.
> I lost several days due to instabilities apparently in Oracle JDK
> 1.6.0_23, which
> caused about 30 junit test failures in contrib (streaming, schedulers, and
> gridmix).
> Switching to JDK 1.6.0_26 made the problems go away.
>
> The release artifacts in this hadoop-1.0.0-rc2 were build with
> JDK 1.6.0_26.
>
> Thank you,
> --Matt (Release Manager for 1.0.0, formerly known as 0.20.205.1)
>
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Roman Shaposhnik <rv...@apache.org>.
On Thu, Dec 8, 2011 at 11:57 AM, Matt Foley <ma...@apache.org> wrote:
> Hello all,
> I have posted a release candidate for Hadoop 1.0.0 at
> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>
> Please download, evaluate, and vote on this list.
> The artifacts will be posted to the maven repo shortly.
> The vote will close at noon PST on Thursday 15 Dec.
Pulled this RC into Bigtop, built the complete stack and published
packages in the usual place (feel free to install on your own cluster
by just doing yum or puppet installs):
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-trunk-rc-hadoop-1.0.0-hive-0.8.0/label=centos5/lastSuccessfulBuild/artifact/output/
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-trunk-rc-hadoop-1.0.0-hive-0.8.0/
Ran test on a fully distributed 5 nodes cluster on EC2 for 10 hours:
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-rc-smoketest/lastCompletedBuild/#showFailuresLink
The bulk of the failures is in Hive.
At this point I feel like we've got to resolve 3 issues with this release:
1. Making sure we agree on what to do with *all* of existing Hive
deployemnts
that are going to be broken by the version change.
2. Analyzing Mahout 0.5 failures:
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-rc-smoketest/lastCompletedBuild/testReport/junit/org.apache.bigtop.itest.mahout.smoke/TestMahoutExamples/factorizeMovieLensRatings/
http://bigtop01.cloudera.org:8080/view/RCs/job/Bigtop-rc-smoketest/lastCompletedBuild/testReport/junit/org.apache.bigtop.itest.mahout.smoke/TestMahoutExamples/testReutersLDA/
3. Considering a fix for multifilewc
Hope this helps.
Thanks,
Roman.
Re: [VOTE] Hadoop-1.0.0 release candidate 2
Posted by Mark Holderbaugh <ho...@yahoo-inc.com>.
Matt,
I would like to request the following JIRA be included in Hadoop 1.0.0.
https://issues.apache.org/jira/browse/MAPREDUCE-3475
https://issues.apache.org/jira/browse/HDFS-2589
Thanks
Mark
On 12/8/11 1:57 PM, "Matt Foley" <ma...@apache.org> wrote:
> Hello all,
> I have posted a release candidate for Hadoop 1.0.0 at
> http://people.apache.org/~mattf/hadoop-1.0.0-rc2/
>
> Please download, evaluate, and vote on this list.
> The artifacts will be posted to the maven repo shortly.
> The vote will close at noon PST on Thursday 15 Dec.
>
> It may be of interest that it is important to use an appropriate version of
> Java.
> I lost several days due to instabilities apparently in Oracle JDK 1.6.0_23,
> which
> caused about 30 junit test failures in contrib (streaming, schedulers, and
> gridmix).
> Switching to JDK 1.6.0_26 made the problems go away.
>
> The release artifacts in this hadoop-1.0.0-rc2 were build with JDK 1.6.0_26.
>
> Thank you,
> --Matt (Release Manager for 1.0.0, formerly known as 0.20.205.1)