That test doesn't do a flush during the increments so doesn't trigger the bug.
Jon. On Wed, Dec 4, 2013 at 8:58 AM, lars hofhansl <[email protected]> wrote: > Interesting that at least the concurrent flush part wasn't caught by the > unit tests (TestAtomicOperation specifically). > > > > ________________________________ > From: Jonathan Hsieh <[email protected]> > To: "[email protected]" <[email protected]> > Sent: Wednesday, December 4, 2013 8:16 AM > Subject: Re: ANN: The second hbase-0.96-1 release candidate is available > for download > > > I've filed an issue for this and will post more results there. (against > 0.96.0 etc). > > https://issues.apache.org/jira/browse/HBASE-10079 > > > On Wed, Dec 4, 2013 at 8:10 AM, Jonathan Hsieh <[email protected]> wrote: > > > -1 > > > > I have a new test I tried on 0.96.1rc1. (polishing to share). Run #2 > with > > flushes is a correctness regression (HBASE-6195 fixed this in 94, and I > had > > verified there but it is broken here).. I'm not sure if #3 (kill) or #4 > > (kill -9) are a regression against 0.94. Here's a link to the testcode > > (its got all kinds of caveats at the moment and will likely get turn into > > integration test). > > > > https://github.com/jmhsieh/hbase/tree/increval > > > > 5 nodes > > hadoop 2.1.0 (a pre cdh5b1 hdfs). > > hbase 0.96.1 rc1 > > > > Test: 250000 increments on a single row an single col with various number > > of client threads (IncrementBlaster). Verify we have a count of 250000 > > after the run (IncrementVerifier). > > > > Run 1: No fault injection. 5 runs. count = 250000. on multiple runs. > > Correctness verified. 1638 inc/s throughput. > > Run 2: flushes table with incrementing row. count = 246875 !=250000. > > correctness failed. 1517 inc/s throughput. > > Run 3: kill of rs hosting incremented row. count = 243750 != 250000. > > Correctness failed. 1451 inc/s throughput. > > Run 4: one kill -9 of rs hosting incremented row. 246878.!= 250000. > > Correctness failed. 1395 inc/s (including recovery) > > > > > > > > On Wed, Nov 27, 2013 at 7:57 PM, Stack <[email protected]> wrote: > > > >> The second hbase-0.96.1 release candidate is available for download: > >> > >> http://people.apache.org/~stack/hbase-0.96.1RC1/ > >> > >> The maven artifacts are available here: > >> > >> https://repository.apache.org/content/repositories/orgapachehbase-016/ > >> > >> The first RC was sunk by: > >> > >> HBASE-10053 ProtobufUtil#toMutationNoData interface change causes > >> downstream > >> > >> We broke hive building against HBase. > >> > >> 137 issues [1] have been fixed since 0.96.0, mostly performance > >> improvement > >> and bug fixes. > >> > >> Shall we release this as hbase-0.96.1? Please vote by December 3rd. > >> > >> Thanks, > >> Your HBase Team > >> > >> 1. http://goo.gl/ZgYBW2 > >> > >> > >> On Tue, Nov 26, 2013 at 4:48 PM, Stack <[email protected]> wrote: > >> > >> > The first hbase-0.96.1 release candidate is available for download: > >> > > >> > http://people.apache.org/~stack/hbase-0.96.1RC0/ > >> > > >> > The staged mvn artifacts can be found here: > >> > > >> > > >> https://repository.apache.org/content/repositories/orgapachehbase-012/ > >> > > >> > 127 issues [1] have been fixed since 0.96.0, mostly performance > >> improvement > >> > and bug fixes. See below for the full list. > >> > > >> > Shall we release this as hbase-0.96.1? Please vote by December 3rd. > >> > > >> > Thanks, > >> > Your HBase Team > >> > > >> > 1. http://goo.gl/ZgYBW2 > >> > > >> > Issues fixed in this RC: > >> > > >> > HBASE-8143 HBase on Hadoop 2 with local short circuit reads (ssr) > causes > >> > OOM > >> > HBASE-10015 Replace intrinsic locking with explicit locks in > >> StoreScanner > >> > HBASE-10026 HBaseAdmin#createTable could fail if region splits too > fast > >> > HBASE-10024 Add an interface to create put with immutable arrays > >> > HBASE-8465 Auto-drop rollback snapshot for snapshot restore > >> > HBASE-10011 Fix some findbugs in the client > >> > HBASE-9865 Reused WALEdits in replication may cause RegionServers to > go > >> OOM > >> > HBASE-9890 MR jobs are not working if started by a delegated user > >> > HBASE-9975 Not starting ReplicationSink when using custom > implementation > >> > for the ReplicationSink. > >> > HBASE-9831 hbasefsck.numthreads' property isn't passed to hbck via > >> cmdline > >> > -D option > >> > HBASE-9963 Remove the ReentrantReadWriteLock in the MemStore > >> > HBASE-9956 Remove keyLength cache from KeyValue > >> > HBASE-9165 HBASE-8084 Improvements to addDependencyJars > >> > HBASE-8084 Improvements to addDependencyJars > >> > HBASE-9138 getHaseIntegrationTestingUtility() is misspelled > >> > HBASE-9970 HBase BulkLoad, table is creating with the timestamp key > also > >> > as a column to the table. > >> > HBASE-9952 Snapshot restore may fail due to NullPointerException > >> > HBASE-9906 Restore snapshot fails to restore the meta edits > sporadically > >> > HBASE-8438 Extend bin/hbase to print a "mapreduce classpath" > >> > HBASE-9915 Performance: isSeeked() in EncodedScannerV2 always returns > >> false > >> > HBASE-8556 fix coverage org.apache.hadoop.hbase.metrics.histogram > >> > HBASE-9872 ModifyTable does not modify the attributes of a newly > >> > modified/changed ColumnDescriptor > >> > HBASE-8559 increase unit-test coverage of package > >> > org.apache.hadoop.hbase.coprocessor > >> > HBASE-8543 fix coverage org.apache.hadoop.hbase.rest.client > >> > HBASE-9902 Region Server is starting normally even if clock skew is > more > >> > than default 30 seconds(or any configured). -> Regionserver node time > is > >> > greater than master node time > >> > HBASE-9849 [REST] Forbidden schema delete in read only mode > >> > HBASE-8552 fix coverage org.apache.hadoop.hbase.rest.filter > >> > HBASE-9850 Issues with UI for table compact/split operation > completion. > >> > After split/compaction operation using UI, the page is not > automatically > >> > redirecting back using IE8/Firefox. > >> > HBASE-4654 [replication] Add a check to make sure we don't replicate > to > >> > ourselves > >> > HBASE-10022 Fix the HE_EQUALS_USE_HASHCODE warning under > >> > hbase-hadoop1-compat > >> > HBASE-9976 Don't create duplicated TableName objects > >> > HBASE-9992 [hbck] Refactor so that arbitrary -D cmdline options are > >> > included > >> > HBASE-5945 Reduce buffer copies in IPC server response path > >> > HBASE-9322 Contention on the SoftValueSortedMap with multiple clients > >> > HBASE-10007 PerformanceEvaluation: Add sampling and latency collection > >> to > >> > randomRead test > >> > HBASE-9989 Add a test on get in TestClientNoCluster > >> > HBASE-10014 HRegion#doMiniBatchMutation rollbacks the memstore even if > >> > there is nothing to rollback. > >> > HBASE-10013 Enhance the message about ZK when there is an exception > >> > HBASE-10001 Add a coprocessor to help testing the performances without > >> > taking into account the i/o > >> > HBASE-10012 Hide ServerName constructor > >> > HBASE-9961 [WINDOWS] Multicast should bind to local address > >> > HBASE-9893 Incorrect assert condition in OrderedBytes decoding > >> > HBASE-9998 Fix javadoc warnings induced by commits > >> > HBASE-9959 Remove some array copy - server side > >> > HBASE-9988 DOn't use HRI#getEncodedName in the client > >> > HBASE-9995 Not stopping ReplicationSink when using custom > implementation > >> > for the ReplicationSink > >> > HBASE-9924 Avoid potential filename conflict in region_mover.rb > >> > HBASE-9982 TestClientNoCluster should use random numbers > >> > HBASE-9973 [ACL]: Users with 'Admin' ACL permission will lose > >> permissions > >> > after upgrade to 0.96.x from 0.94.x or 0.92.x > >> > HBASE-9987 Remove some synchronisation points in HConnectionManager > >> > HBASE-9985 Add HBase shell command to start/stop tracing > >> > HBASE-9983 Lower the memory footprint of HRegionLocation > >> > HBASE-9869 Optimize HConnectionManager#getCachedLocation > >> > HBASE-9954 Incorporate HTTPS support for HBase > >> > HBASE-9908 [WINDOWS] Fix filesystem / classloader related unit tests > >> > HBASE-9112 Custom TableInputFormat in initTableMapperJob throws > >> > ClassNoFoundException on TableMapper > >> > HBASE-9947 Add CM action for online compression algorithm change > >> > HBASE-9958 Remove some array copy, change lock scope in locateRegion > >> > HBASE-9880 client.TestAsyncProcess.testWithNoClearOnFail broke on 0.96 > >> by > >> > HBASE-9867 > >> > HBASE-9810 Global memstore size will be calculated wrongly if > replaying > >> > recovered edits throws exception > >> > HBASE-9907 HBASE-9775 Rig to fake a cluster so can profile client > >> behaviors > >> > HBASE-9870 HFileDataBlockEncoderImpl#diskToCacheFormat uses wrong > format > >> > HBASE-9757 Reenable fast region move in SlowDeterministicMonkey > >> > HBASE-9710 Use the region name, not the encoded name, when region is > not > >> > on current server > >> > HBASE-9926 Scanner doesn't check if a region is available > >> > HBASE-9918 MasterAddressTracker & ZKNamespaceManager ZK listeners are > >> > missed after master recovery > >> > HBASE-9917 HBASE-9775 Fix it so Default Connection Pool does not spin > up > >> > max threads even when not needed > >> > HBASE-9895 0.96 Import utility can't import an exported file from 0.94 > >> > HBASE-9957 [WINDOWS] TestNamespaceUpgrade fails on windows > >> > HBASE-9885 Avoid some Result creation in protobuf conversions > >> > HBASE-9563 Autorestart doesn't work if zkcleaner fails > >> > HBASE-9792 Region states should update last assignments when a region > is > >> > opened. > >> > HBASE-9813 Log splitting doesn't prevent RS creating new hlog file > >> > HBASE-9900 Fix unintended byte[].toString in AccessController > >> > HBASE-9909 TestHFilePerformance should not be a unit test, but a tool > >> > HBASE-9737 Corrupt HFile cause resource leak leading to Region Server > >> OOM > >> > HBASE-9751 Excessive readpoints checks in StoreFileScanner > >> > HBASE-9716 LoadTestTool should provide default min and max settings to > >> the > >> > data generator > >> > HBASE-7600 TestAdmin.testCreateBadTables is failing occasionally > >> > HBASE-9807 block encoder unnecessarily copies the key for each reseek > >> > HBASE-9753 Excessive readpoint checks in MemstoreScanner > >> > HBASE-9745 HBASE_CLASSPATH to end of Java classpath and use another > env > >> > var for prefix > >> > HBASE-9851 TestHBaseFsck.testQuarantineMissingHFile is flaky > >> > HBASE-9747 PrefixFilter with OR condition gives wrong results > >> > HBASE-9727 HBase Rest Server - DELETE scanner operation is a no-op > >> > HBASE-9221 Provide interface for getting a User in the client > >> > HBASE-9886 Optimize ServerName#compareTo > >> > HBASE-9837 HBASE-9080 'Retain assignment should be used when > re-enabling > >> > table(s)' > >> > HBASE-9901 Add a toString in HTable, fix a log in AssignmentManager > >> > HBASE-9855 evictBlocksByHfileName improvement for bucket cache > >> > HBASE-9836 Intermittent > >> > > >> > TestRegionObserverScannerOpenHook#testRegionObserverCompactionTimeStacking > >> > failure > >> > HBASE-9859 Canary Shouldn't go off if the table being read from is > >> disabled > >> > HBASE-9867 Save on array copies with a subclass of LiteralByteString > >> > HBASE-9868 Remove some array copy, especially around protobuf > >> > HBASE-9871 PB issue : Increment attributes are not getting passed to > >> server > >> > HBASE-9593 Region server left in online servers list forever if it > went > >> > down after registering to master and before creating ephemeral node > >> > HBASE-9862 manage error per server and per region in the protobuffed > >> client > >> > HBASE-9861 Location does not have to be refreshed on regionTooBusy > >> > HBASE-9659 some integration tests can no longer be run using maven > >> > HBASE-9822 IntegrationTestLazyCfLoading failed occasionally in a > secure > >> > enviroment > >> > HBASE-9843 Various fixes in client code > >> > HBASE-9750 Add retries around Action server stop/start > >> > HBASE-9759 Prevent random number collision in IntegrationTestBulkLoad > >> > HBASE-9848 Make regionserver Canary exit with different exit codes > >> > HBASE-9765 TestImportTsv.testBulkOutputWithTsvImporterTextMapper() may > >> not > >> > be working correctly > >> > HBASE-7525 A canary monitoring program specifically for regionserver > >> > HBASE-9748 HBASE-9696 > >> > HBASE-9827 Intermittent > TestLogRollingNoCluster#testContendedLogRolling > >> > failure > >> > HBASE-9785 Fix heap size reporting in HRegion > >> > HBASE-9821 Scanner id could collide > >> > HBASE-9694 HBaseTestingUtility.createTable infinite loop > >> > HBASE-9772 Normalize new client default values > >> > HBASE-9768 Two issues in AsyncProcess > >> > HBASE-9784 Switch to Hadoop-2.2 > >> > HBASE-9793 Offline a region before it's closed could cause double > >> > assignment > >> > HBASE-9798 Include dependency hamcrest-core > >> > HBASE-9796 npe in RegionServerCallable > >> > HBASE-9776 Test Load And Verify Fails with TableNotEnabledException > >> > HBASE-9788 [WINDOWS] Update rest server class name > >> > HBASE-9755 Cannot run classes in hbase-server tests jar from command > >> line > >> > HBASE-9791 MR initializes scanner twice > >> > HBASE-9773 Master aborted when hbck asked the master to assign a > region > >> > that was already online > >> > HBASE-9720 > >> > TestSplitTransactionOnCluster#testShutdownFixupWhenDaughterHasSplit > >> > occasionally times out > >> > HBASE-9771 [WebUI] Humanize store and blockcache statistics on RS > >> > HBASE-9570 With AccessDeniedException, HBase shell would be better to > >> just > >> > display the error message to be user friendly > >> > HBASE-9758 Log missing rows and their regions if ITBLL fails > >> > HBASE-9692 TestOrderedBytes doesn't display the arrays correctly > >> > > >> > > >> > > > > > > > > -- > > // Jonathan Hsieh (shay) > > // Software Engineer, Cloudera > > // [email protected] > > > > > > > > > -- > // Jonathan Hsieh (shay) > // Software Engineer, Cloudera > // [email protected] > -- // Jonathan Hsieh (shay) // Software Engineer, Cloudera // [email protected]
