[HBase-dev] ANN: The first hbase 0.94.1 release candidate is available for download

I created a public Google spreadsheet: https://docs.google.com/spreadsheet/ccc?key=0AvLqcVIqNtlTdHM0LTlwQVAyUDJ2U2Z1bEprSEYxdXc#gid=0

If you do some testing on the 0.94.1RC please put a note there. Any testing help is greatly appreciated.

Feel free to add new tests there as well as you see fit.

This is an experiment of course. If I find that document vandalized I'll need to remove it.

Thanks.

-- Lars

Lars
I think you mean 'whether we should release this as 0.94.1'?

Regards
Ram

Ah, yes. Indeed :)

Working on security tests with LoadTestTool now.

Best regards,

- Andy

Problems worthy of attack prove their worth by hitting back. - Piet
Hein (via Tom White)

Currently I'm -0.5 because tests don't pass fully for me locally --
TestReplication hangs with the same jstack as attached to HBASE-6406
as https://issues.apache.org/jira/secure/attachment/12536942/testReplication.jstack
.

Best regards,

- Andy

Problems worthy of attack prove their worth by hitting back. - Piet
Hein (via Tom White)

Here is my comment about consistently hanging TestZooKeeper:

https://issues.apache.org/jira/browse/HBASE-6389?focusedCommentId=13417697&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13417697

If that test consistently fails the RC should be -1'ed.
I'll double check later tonight.

I opened https://issues.apache.org/jira/browse/HBASE-6424, trying to
get to it today but maybe not.

I am hereby sinking RC0. That was quick.
We had too many failed/hung tests. Turns out the culprit wasHBASE-6389, which has been reverted.


If you commit against 0.94, please retarget to 0.94.1.


Stay tuned for RC1.


Thanks.

-- Lars
________________________________
From: lars hofhansl <lhofhansl@yahoo.com>
To: Andrew Purtell <apurtell@apache.org>; "dev@hbase.apache.org" <dev@hbase.apache.org>
Sent: Wednesday, July 18, 2012 4:22 PM
Subject: Re: ANN: The first hbase 0.94.1 release candidate is available for download

If that test consistently fails the RC should be -1'ed.
I'll double check later tonight.

The 2nd 0.94.1 RC is available for download here: http://people.apache.org/~larsh/hbase-0.94.1-rc1/

Signed with my code signing key: A1AC25A9

0.94.1 is a bug fix release and has 156 issues resolved against it.

The full list of changes is available here:
https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&jqlQuery=project+%3D+HBASE+AND+fixVersion+%3D+%220.94.1%22+AND+status+%3D+Resolved+ORDER+BY+priority+DESC&mode=hide

The tests on the Jenkins machines are stable again and the artifact were built there.

Please take the RC for a spin, check out the doc, etc, and vote +1/-1 by August 7th on whether we should release this as 0.94.1.

The testing spreadsheet is here:
https://docs.google.com/spreadsheet/ccc?key=0AvLqcVIqNtlTdDAxUmwzVkpXSl9DN0tGUmYxUFhub0E#gid=0

Thanks.

-- Lars

On taking it for a spin I'm having issues when I kill the RS and DN that's
holding root.

Spun up a new dfs on hadoop-0.20.2-cdh3u3
Started hbase
started running loadtest tool.
killed rs and dn holding root with killall -9 java on server sv4r27s44 at
about 2012-07-25 22:40:00

After things stabilize Root is in a bad state. Ran hbck and got:
Exception in thread "main"
org.apache.hadoop.hbase.client.NoServerForRegionException: No server
address listed in -ROOT- for region .META.,,1.1028785192 containing row
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegionInMeta(HConnectionManager.java:1016)
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegion(HConnectionManager.java:841)
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegion(HConnectionManager.java:810)
at org.apache.hadoop.hbase.client.HTable.finishSetup(HTable.java:232)
at org.apache.hadoop.hbase.client.HTable.(HBaseFsck.java:241)
at org.apache.hadoop.hbase.util.HBaseFsck.main(HBaseFsck.java:3236)



hbase(main):001:0> scan '-ROOT-'
ROW COLUMN+CELL


12/07/25 22:43:18 INFO security.UserGroupInformation: JAAS Configuration
already set up for Hadoop, not re-installing.
.META.,,1 column=info:regioninfo,
timestamp=1343255838525, value={NAME => '.META.,,1', STARTKEY => '', ENDKEY
=> '', ENCODED => 1028785192,}
.META.,,1 column=info:v,
timestamp=1343255838525, value=\x00\x00

1 row(s) in 0.5930 seconds


Here's the master log: https://gist.github.com/3179194

I tried the same thing with 0.92.1 and I was able to get into a similar
situation, so I don't think this is anything new. I'm going to file an
issue with what I've seen, since all that was required to fix it was a move
of meta.

Thanks Elliot!
I wonder whether this is related to (or the cause of) HBASE-5754?

-- Lars

________________________________
From: Elliott Clark <eclark@stumbleupon.com>
To: dev@hbase.apache.org; lars hofhansl <lhofhansl@yahoo.com>
Sent: Thursday, July 26, 2012 12:39 PM
Subject: Re: ANN: The 2nd hbase 0.94.1 release candidate is available for download

On taking it for a spin I'm having issues when I kill the RS and DN that's
holding root.

Spun up a new dfs on hadoop-0.20.2-cdh3u3
Started hbase
started running loadtest tool.
killed rs and dn holding root with killall -9 java on server sv4r27s44 at
about 2012-07-25 22:40:00

After things stabilize Root is in a bad state. Ran hbck and got:
Exception in thread "main"
org.apache.hadoop.hbase.client.NoServerForRegionException: No server
address listed in -ROOT- for region .META.,,1.1028785192 containing row
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegionInMeta(HConnectionManager.java:1016)
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegion(HConnectionManager.java:841)
at
org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.locateRegion(HConnectionManager.java:810)
at org.apache.hadoop.hbase.client.HTable.finishSetup(HTable.java:232)
at org.apache.hadoop.hbase.client.HTable.(HBaseFsck.java:241)
at org.apache.hadoop.hbase.util.HBaseFsck.main(HBaseFsck.java:3236)



hbase(main):001:0> scan '-ROOT-'
ROW                                           COLUMN+CELL


12/07/25 22:43:18 INFO security.UserGroupInformation: JAAS Configuration
already set up for Hadoop, not re-installing.
.META.,,1                                    column=info:regioninfo,
timestamp=1343255838525, value={NAME => '.META.,,1', STARTKEY => '', ENDKEY
=> '', ENCODED => 1028785192,}
.META.,,1                                    column=info:v,
timestamp=1343255838525, value=\x00\x00

1 row(s) in 0.5930 seconds


Here's the master log: https://gist.github.com/3179194

I tried the same thing with 0.92.1 and I was able to get into a similar
situation, so I don't think this is anything new.  I'm going to file an
issue with what I've seen, since all that was required to fix it was a move
of meta.

Sounds bad. Its unrelated to HBASE-5754 I'd say (its -ROOT- rather
than .META.).
St.Ack

Seems like a blocker if hbck failure is reproducible even though it
happens with 0.92. We know about it now... Thanks Elliott.

I don't think that this is an hbck issue. The root table is corrupted
before the hbck run. The hbck check just shows that root does not
contain an assigned meta.

Yes but the user can't get the table back online with hbck, no?

There are two issues here, correct me if I'm not on the same page
because I'm missing some information (like what hbck switches you
used).

1) The root table is hosed.

2) hbck craps out with Exception in thread "main"
org.apache.hadoop.hbase.client.NoServerForRegionException instead of
fixing the problem.

That seems to be the case.

0.94.1 should get out of the door soon, though, because of HBASE-6311, hence (and because this exists in 0.92 and probably 0.90)
I don't think this should be blocker for 0.94.1.

Could do a small 0.94.2 release to address this issue afterwards.


Maybe we have to start thinking about a.b.c.d bug fix releases to get "emergency" fixes out (0.94.0.1 or 0.94.1.1 in this case).

-- Lars

+1 on the above from Lars based on the additional comment on HBASE-6461

HBCK should be able to get the cluster back in this state, that can be
for 0.94.2.

Best regards,

- Andy

Problems worthy of attack prove their worth by hitting back. - Piet
Hein (via Tom White)

Here is my (non-binding) +1 based on below:

- Checked signatures
- Run some basic smoke tests
- Run LoadTestTool, TestAcidGuarantees on a 7 node cluster
- Ported the patch for HBASE-6241 (v3 patch, which is coming soon), and
which runs LoadTestTool while killing the servers, rolling restart, batch
killing 50%, etc.

Minor issues (should not sink RC)
- no md5 or sha file
- run into HBASE-6469

Enis

On more testing, I'm (a non-binding) +1.
Tested on hadoop 2.0.0-alpha and cloudera's cdh3 using Loadtest Tool and
acid verification.

I am also +1 (non-binding). I ran some medium sized tests like TestLoadAndVerify (Bigtop's), and they ran fine.
Devaraj.

+1 (non-binidng)

We have ran few basic scenarios with some 2TB data, 1500 regions in a 5 node
cluster.

It will be good in future if we share hbase tar ball with hadoop
2.0.0-alpha, so that we can test the release in both the hadoop version.

Thanks & Regards,
Gopinathan A

I'd say nah. Lets keep our versioning basic. Instead lets work on
making releases less of a pain (smile).

Good on you Lars,
St.Ack

This is a reminder that the voting phase for this RC ends today.
We have a few +1 and no -1. If I hear no objections by this evening I will release


As for myself, I'm +1 as well:
I have been playing with HBase 0.94.1 in various incarnations (local, on top of Hadoop 1.0.3 and 2.0.0-alpha),
did created tables, did scans/puts/gets, tested out some coprocessors, etc.

-- Lars

+1

I would have liked to try it on our prod clusters but we only got so
far as our dev cluster.

J-D

+1

Downloaded it. Started a standalone instance. Loaded up a couple
million of rows. Verified count. Checked logs for errors. Looked
good except for the below which is not enough to sink a release.

Docs look fine.

Tarball should have an md5 published alongside it -- see the hadoop
how to release doc.

St.Ack



2012-08-07 16:58:43,226 INFO
org.apache.zookeeper.server.ZooKeeperServer: Client attempting to
establish new session at /127.0.0.1:60142
2012-08-07 16:58:43,232 INFO
org.apache.zookeeper.server.ZooKeeperServer: Established session
0x13901f7792d0009 with negotiated timeout 40000 for client
/127.0.0.1:60142
2012-08-07 16:59:16,777 WARN
org.apache.zookeeper.server.NIOServerCnxn: caught end of stream
exception
EndOfStreamException: Unable to read additional data from client
sessionid 0x13901f7792d0009, likely client has closed socket
at org.apache.zookeeper.server.NIOServerCnxn.doIO(NIOServerCnxn.java:220)
at org.apache.zookeeper.server.NIOServerCnxnFactory.run(NIOServerCnxnFactory.java:224)
at java.lang.Thread.run(Thread.java:662)

Non-binding +1

Installed in a 5 node cluster, ran some standard and some custom load
tests. No issues so far.

Regards,
Aditya...

+1 from me and my team. Here in Huawei our testing team has tested some scenarios and the same has been updated in the excel sheet provided by Lars.

Our testing has been done on hadoop 0.20.205.

Regards
Ram