hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jonathan Hsieh (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-5458) Thread safety issues with Compression.Algorithm.GZ and CompressionTest
Date Thu, 17 Jan 2013 20:02:18 GMT

    [ https://issues.apache.org/jira/browse/HBASE-5458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13556537#comment-13556537
] 

Jonathan Hsieh commented on HBASE-5458:
---------------------------------------

[~eclark] lgtm +1.

[~dmcintosh] it looks good to me -- this idiom is called double-checked locking and is an
optimization for lazy initializers.  In your example let's say thread 1 is already in buildCodec.
 This means it has the lock.   thread 2 could see null, and would get blocked by the lock.
 thread 1 would finish initialization, set the variable, and then release the lock.  thread
2 enters, sees that the variable is not null and falls through.  I don't see where the conf
object gets passed elsewhere.  Am I missing something?
                
> Thread safety issues with Compression.Algorithm.GZ and CompressionTest
> ----------------------------------------------------------------------
>
>                 Key: HBASE-5458
>                 URL: https://issues.apache.org/jira/browse/HBASE-5458
>             Project: HBase
>          Issue Type: Bug
>          Components: io
>    Affects Versions: 0.90.5
>            Reporter: David McIntosh
>            Assignee: Elliott Clark
>            Priority: Minor
>         Attachments: HBASE-5458-090-0.patch, HBASE-5458-090-1.patch, HBASE-5458-090-2.patch
>
>
> I've seen some occasional NullPointerExceptions in ZlibFactory.isNativeZlibLoaded(conf)
during region server startups and the completebulkload process.  This is being caused by a
null configuration getting passed to the isNativeZlibLoaded method.  I think this happens
when 2 or more threads call the CompressionTest.testCompression method at once.  If the GZ
algorithm has not been tested yet both threads could continue on and attempt to load the compressor.
 For GZ the getCodec method is not thread safe which could lead to one thread getting a reference
to a GzipCodec that has a null configuration.
> {code}
> current:
>       DefaultCodec getCodec(Configuration conf) {
>         if (codec == null) {
>           codec = new GzipCodec();
>           codec.setConf(new Configuration(conf));
>         }
>         return codec;
>       }
> {code}
> one possible fix would be something like this:
> {code}
>       DefaultCodec getCodec(Configuration conf) {
>         if (codec == null) {
>           GzipCodec gzip = new GzipCodec();
>           gzip.setConf(new Configuration(conf));
>           codec = gzip;
>         }
>         return codec;
>       }
> {code}
> But that may not be totally safe without some synchronization.  An upstream fix in CompressionTest
could also prevent multi thread access to GZ.getCodec(conf)
> exceptions:
> 12/02/21 16:11:56 ERROR handler.OpenRegionHandler: Failed open of region=all-monthly,,1326263896983.bf574519a95263ec23a2bad9f5b8cbf4.
> java.io.IOException: java.lang.NullPointerException
>         at org.apache.hadoop.hbase.util.CompressionTest.testCompression(CompressionTest.java:89)
>         at org.apache.hadoop.hbase.regionserver.HRegion.checkCompressionCodecs(HRegion.java:2670)
>         at org.apache.hadoop.hbase.regionserver.HRegion.openHRegion(HRegion.java:2659)
>         at org.apache.hadoop.hbase.regionserver.HRegion.openHRegion(HRegion.java:2647)
>         at org.apache.hadoop.hbase.regionserver.handler.OpenRegionHandler.openRegion(OpenRegionHandler.java:312)
>         at org.apache.hadoop.hbase.regionserver.handler.OpenRegionHandler.process(OpenRegionHandler.java:99)
>         at org.apache.hadoop.hbase.executor.EventHandler.run(EventHandler.java:158)
>         at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
>         at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
>         at java.lang.Thread.run(Thread.java:662)
> Caused by: java.lang.NullPointerException
>         at org.apache.hadoop.io.compress.zlib.ZlibFactory.isNativeZlibLoaded(ZlibFactory.java:63)
>         at org.apache.hadoop.io.compress.GzipCodec.getCompressorType(GzipCodec.java:166)
>         at org.apache.hadoop.io.compress.CodecPool.getCompressor(CodecPool.java:100)
>         at org.apache.hadoop.io.compress.CodecPool.getCompressor(CodecPool.java:112)
>         at org.apache.hadoop.hbase.io.hfile.Compression$Algorithm.getCompressor(Compression.java:236)
>         at org.apache.hadoop.hbase.util.CompressionTest.testCompression(CompressionTest.java:84)
>         ... 9 more
> Caused by: java.io.IOException: java.lang.NullPointerException
>         at org.apache.hadoop.hbase.util.CompressionTest.testCompression(CompressionTest.java:89)
>         at org.apache.hadoop.hbase.io.hfile.HFile$Reader.readTrailer(HFile.java:890)
>         at org.apache.hadoop.hbase.io.hfile.HFile$Reader.loadFileInfo(HFile.java:819)
>         at org.apache.hadoop.hbase.mapreduce.LoadIncrementalHFiles.groupOrSplit(LoadIncrementalHFiles.java:405)
>         at org.apache.hadoop.hbase.mapreduce.LoadIncrementalHFiles$2.call(LoadIncrementalHFiles.java:323)
>         at org.apache.hadoop.hbase.mapreduce.LoadIncrementalHFiles$2.call(LoadIncrementalHFiles.java:321)
>         at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
>         at java.util.concurrent.FutureTask.run(FutureTask.java:138)
>         at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
>         at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
>         at java.lang.Thread.run(Thread.java:662)
> Caused by: java.lang.NullPointerException
>         at org.apache.hadoop.io.compress.zlib.ZlibFactory.isNativeZlibLoaded(ZlibFactory.java:63)
>         at org.apache.hadoop.io.compress.GzipCodec.getCompressorType(GzipCodec.java:166)
>         at org.apache.hadoop.io.compress.CodecPool.getCompressor(CodecPool.java:100)
>         at org.apache.hadoop.io.compress.CodecPool.getCompressor(CodecPool.java:112)
>         at org.apache.hadoop.hbase.io.hfile.Compression$Algorithm.getCompressor(Compression.java:236)
>         at org.apache.hadoop.hbase.util.CompressionTest.testCompression(CompressionTest.java:84)
>         ... 10 more

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message