Thanks for bringing this up, Lars. I looked at three builds where this test OOM'ed: 631, 634 and 635.
It seems that this problem is reproducible on ubuntu3. I wonder if juggling of build servers would make this OOM disappear. Cheers On Mon, Dec 17, 2012 at 9:58 PM, lars hofhansl <[email protected]> wrote: > Does anybody know what may have caused the recent OOM failures in > TestHFileBlock.testConcurrentReading[1]? > > > This is the exception: > > > Caused by: java.lang.OutOfMemoryError > at java.util.zip.Inflater.init(Native Method) > at java.util.zip.Inflater.<init>(Inflater.java:83) > at > org.apache.hadoop.io.compress.zlib.BuiltInGzipDecompressor.<init>(BuiltInGzipDecompressor.java:45) > at > org.apache.hadoop.io.compress.GzipCodec.createDecompressor(GzipCodec.java:136) > at > org.apache.hadoop.io.compress.CodecPool.getDecompressor(CodecPool.java:127) > at > org.apache.hadoop.hbase.io.hfile.Compression$Algorithm.getDecompressor(Compression.java:290) > at > org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader.decompress(HFileBlock.java:1397) > at > org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderV2.readBlockDataInternal(HFileBlock.java:1830) > at > org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderV2.readBlockData(HFileBlock.java:1643) > at > org.apache.hadoop.hbase.io.hfile.TestHFileBlock$BlockReaderThread.call(TestHFileBlock.java:639) > at > org.apache.hadoop.hbase.io.hfile.TestHFileBlock$BlockReaderThread.call(TestHFileBlock.java:603) > at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303) > at java.util.concurrent.FutureTask.run(FutureTask.java:138) > > > Here's the latest test run with that failure: > https://builds.apache.org/job/HBase-0.94/635/ > > Looks like this is creating a new Decompressor for each single block. > Looking at the code that seems to be by design when the > BuiltInGzipDecompressor is used. > Seems somewhat inefficient, though. > > > I initially thought this was caused by HBASE-7336, but that turned out to > be not the case (OOMs still occurred with that change reverted). > > If anybody knows anything about this, please let me know. It might also > just be an environment issue. > > > Thanks. > > > -- Lars > >
