Trying this patch to see if it cures hudson of leakage
M src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java
Add a passedZkCluster flag. If we didn't start the zk cluster,
don't shut it down on way out.
M src/test/java/org/apache/hadoop/hbase/TestHBaseTestingUtility.java
Add a test that starts up three clusters in a vm all sharing single
zk, each to its own chroot location. Add to tables in each and
very the add by doing same as the old TestMultiClusters used)
M src/test/java/org/apache/hadoop/hbase/replication/TestReplication.java
Reenable shutdown of started clusters. Hopefully this clears
the leaking of zk ensembles up on hudson.
M src/main/java/org/apache/hadoop/hbase/zookeeper/ZooKeeperWatcher.java
Testing, it looks like legitimately the callback for session connected
can be invoked before construction of zookeeperwatcher finishes. In
particular the zookeeper data member can be null. Hang around for a
second of so before throwing an exception (Make the exception
indicate the particular zkw by adding to the error message stack
trace made at zkw construction -- helps with debugging this stuff)
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1023181 13f79535-47bb-0310-9956-ffa450edef68
M src/main/java/org/apache/hadoop/hbase/master/ServerManager.java
Bit of javadoc.
M src/main/java/org/apache/hadoop/hbase/master/AssignmentManager.java
If bad server, pass in name of bad server when remaking assignment
plan -- i missed this important part on commit of part 2 of 3113.
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1023179 13f79535-47bb-0310-9956-ffa450edef68
M src/main/java/org/apache/hadoop/hbase/zookeeper/ZKUtil.java
Changed mentions of 'quorum' to 'ensemble'.
M src/main/java/org/apache/hadoop/hbase/mapreduce/IdentityTableReducer.java
M src/main/java/org/apache/hadoop/hbase/mapreduce/TableInputFormat.java
Minor formatting.
M src/main/java/org/apache/hadoop/hbase/mapreduce/TableInputFormatBase.java
Removed unused imports and minor formatting.
M src/main/java/org/apache/hadoop/hbase/mapreduce/TableMapReduceUtil.java
Documented what the quorumAddress parameter is for.
Removed an unnecessary looking HBaseConfiguration.addHbaseResources(conf);
(and adjusted documentation of job to say no hbase config set by the
reduce setup method).
M src/main/java/org/apache/hadoop/hbase/mapreduce/TableOutputFormat.java
Made this class implment Configurable. Moved creation of table from
RecordWrite constructor based off passed TaskAttemptContext instead
into the new setConf method. Added table and conf data members.
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1022825 13f79535-47bb-0310-9956-ffa450edef68
it fails to replicate
HBASE-3044 [replication] ReplicationSource won't cleanup logs if there's
nothing to replicate
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1022770 13f79535-47bb-0310-9956-ffa450edef68
M src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java
Some fixup to support stepped start up of mini cluster; allow
starting dfs cluster, then later put an hbase mini cluster on top.
(startMiniHBaseCluster, createRootDir): Added.
D src/test/java/org/apache/hadoop/hbase/AbstractMergeTestBase.java
Removed messy subclass of HBaseClusterTestCase used building
up some specific loaded regions. Replaced with utility added
to HBaseTestingUtility and by methods added to specific test.
D src/test/java/org/apache/hadoop/hbase/util/TestMergeMeta.java
Deleted test that did nothing -- test merging of .META. -- but
the superclass was making user regions, not multiple instances
of .META. -- which we don't support anyways currently.
M src/test/java/org/apache/hadoop/hbase/util/TestMergeTable.java
Rewritten to use HBaseTestingUtility. Also added assertions that
it actually did successful merge (Were none previous).
M src/main/java/org/apache/hadoop/hbase/catalog/CatalogTracker.java
Added a new constructor. Are the others redundant given I just
added implementation of Abortable to HConnection interface (the
implmementation of HConnection used implement it -- I've just
moved it up into the Interface itself).
M src/main/java/org/apache/hadoop/hbase/util/HMerge.java
Bit of minor cleanup refactoring.
M src/main/java/org/apache/hadoop/hbase/client/HConnectionManager.java
The HConnection Interface now implements Abortable.
M src/main/java/org/apache/hadoop/hbase/client/HConnection.java
Extend Abortable (The implementation was implementing Abortable
anyways).
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1006219 13f79535-47bb-0310-9956-ffa450edef68
M src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java
Added passing a Configuration to createTable.
M src/test/java/org/apache/hadoop/hbase/HBaseClusterTestCase.java
Make a new configuration each time we create a cluster else this
old testing utility fails when we do restart in middle of a test
suite using this old stuff.
M src/test/java/org/apache/hadoop/hbase/regionserver/TestHRegion.java
Fixed imports.
M src/test/java/org/apache/hadoop/hbase/MiniHBaseCluster.java
Added catch for other than IOE... shutdown if anything thrown.
M src/test/java/org/apache/hadoop/hbase/master/TestCatalogJanitor.java
This test was for sure horked. Fixed it so we were testing
the javadoc description of how its supposed to work.
M src/test/java/org/apache/hadoop/hbase/master/TestDeadServer.java
Another test that just wasn't working -- couldn't have worked on
commit.
M src/test/java/org/apache/hadoop/hbase/master/TestLoadBalancer.java
Must have resolvable host (How did this test ever work)?
M src/test/java/org/apache/hadoop/hbase/client/TestGetRowVersions.java
Fixup to make this test pass. New Configuration after cluster restart.
M src/test/java/org/apache/hadoop/hbase/client/TestScannerTimeout.java
Fixup to make this test pass. The low scanner timeout was stopping
this test from even starting up.
M src/test/java/org/apache/hadoop/hbase/client/TestFromClientSide.java
Use new method from HTU.
M src/main/java/org/apache/hadoop/hbase/regionserver/SplitTransaction.java
If passed server is null, we are in test mode.
M src/main/java/org/apache/hadoop/hbase/catalog/CatalogTracker.java
Catch other dropped connection exception types when trying to prove
we have a good root/meta address.
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1005990 13f79535-47bb-0310-9956-ffa450edef68
HConnections go stale if miniCluster is shutdown on them. Make it
so we make new HConnections on other side of a new cluster spin up.
Was cause for failing of last few tests.
M src/test/java/org/apache/hadoop/hbase/HBaseTestingUtility.java
So if you shutdown a cluster and you have HConnections that were
made using the HTUs Configuration, they'll all be stale --
session expired -- and won't be of any use connecting to new cluster
that comes up... now that HConnection is first class dependent on
its own ZooKeeperWatcher instance.
So, I went through this class and fixed it up so we use new
Configurations when could be called either side of a mini cluster
restart. Added note to getConfiguration to be careful what you
do with it because Configuration makes for an HConnection.
M src/test/java/org/apache/hadoop/hbase/TestZooKeeper.java
Don't use same Configuration everywhere.
M src/test/java/org/apache/hadoop/hbase/master/TestRestartCluster.java
Make new Configuration when we make new Cluster so we get fresh
HConnection -- not one w/ stale data, disconnected.
M src/main/java/org/apache/hadoop/hbase/master/ServerManager.java
Check that new server that comes in on a regionServerReport
doesn't have same port and host as one we have already registered
as we do in reportForDuty call (This is preemtive bug fix).
M src/main/java/org/apache/hadoop/hbase/client/HConnectionManager.java
Save of an identifier for HCM and use it in log messages, particularly
when we close -- helps debugging.
M src/main/java/org/apache/hadoop/hbase/client/HBaseAdmin.java
getConfiguration added.
git-svn-id: https://svn.apache.org/repos/asf/hbase/trunk@1005714 13f79535-47bb-0310-9956-ffa450edef68
Michael Stack
Jean-Daniel Cryans