jimk in lucene

HADOOP-2668 Documentation and improved logging so fact that hbase now requires migration comes as less of a surprise

HADOOP-2643 Make migration tool smarter.
HADOOP-2525 Master changes to prevent looping
HADOOP-2525 Same 2 lines repeated 11 million times in HMaster log upon HMaster shutdown

HADOOP-2616 hbase not spliting when the total size of region reaches max region size * 1.5

HADOOP-2587 Splits blocked by compactions cause region to be offline for duration of compaction.

Patch verified by Billy Pearson

HADOOP-2558 modified CHANGES.txt to indicate that this is an incompatible change.
HADOOP-2558 org.onelab.filter.BloomFilter class uses 8X the memory it should be using

HADOOP-2500 Unreadable region kills region servers

HADOOP-2587 Splits blocked by compactions cause region to be offline for duration of compaction.

Fix bug in TestCompaction in which two mini dfs clusters were being started for the same test.

HADOOP-2478 Restructure how HBase lays out files in the file system (phase 2) Includes migration tool org.apache.hadoop.hbase.util.Migrate

  1. … 22 more files in changeset.
HADOOP-2478 test input data.
  1. /hadoop/trunk/src/contrib/hbase/src/testdata
HADOOP-2468 TestRegionServerExit failed in Hadoop-Nightly #338
HADOOP-2458 HStoreFile.writeSplitInfo should just call HStoreFile.Reference.write

HADOOP-2458 HStoreFile.writeSplitInfo should just call HStoreFile.Reference.write

HADOOP-2451 End key is incorrectly assigned in many region splits
HADOOP-2441 Fix build failures in TestHBaseCluster
HADOOP-2199 Add tools for going from hregion filename to region name in logs
HADOOP-2430 Master will not shut down if there are no active region servers
HADOOP-2414 Fix ArrayIndexOutOfBoundsException in bloom filters.
HADOOP-2418 Fix assertion failures in TestTableMapReduce, TestTableIndex, and TestTableJoinMapReduce

HADOOP-2417 Fix critical shutdown problem introduced by HADOOP-2338

HADOOP-2392, HADOOP-2324:


- initialChore() now returns boolean


- rather than retry in root and meta scanners, return if a scan fails. It will get retried on the next scan. This has two effects: 1) scanners exit more quickly during shutdown and 2) they don't keep retrying to connect to a dead server, allowing them to recover from a server going down more quickly.

- initialScan in root and meta scanners return boolean and do not progress to maintenanceScan until the initial scan completes successfully.


- speed up region server exit by reordering join's so that we join with threads in the order that we told them to stop


- remove overrides of heartbeat and thread wake intervals



- move check for null HRegionInfo before first attempt to dereference it.


- HMaster$BaseScanner.checkAssigned: don't try to split dead server's log if initial startup has completed.



- change toString() to only output the region name rather than calling HRegionInfo.toString()


- make logging a bit less verbose


- was writing to local file system and failing on Windows

HADOOP-2392 TestRegionServerExit has new failure mode since HADOOP-2338
HADOOP-2350 Scanner api returns null row names, or skips row names if different column families do not have entries for some rows
HADOOP-2338 Fix NullPointerException in master server.
HADOOP-2365 Result of HashFunction.hash() contains all identical values
HADOOP-2333 Client side retries happen at the wrong level
HADOOP-2321 TestScanner2 does not release resources which sometimes cause the test to time out
HADOOP-2309 ConcurrentModificationException doing get of all region start keys
HADOOP-2295 Fix assigning a region to multiple servers
HADOOP-2261 HTable.abort no longer throws exception if there is no active update.

HADOOP-2287 Make hbase unit tests take less time to complete.

HADOOP-2262 Retry n times instead of n**2 times.

HADOOP-2262 HADOOP-2261 fail fast on non-existing table, change abort to function after commit even if commit was successful