Are there any other threads running? Can you post their stack traces too? Are you sure nothing is happening? EG, if you look in the index, do you see files slowly increasing in size (indicating there is a merge running).
These two traces are actually normal. The ArticleIngestor thread is trying to close, but will not return until all BG merges have completed. It calls IndexWriter.wait, which releases the lock and waits for the BG merges to notify completion. The Lucene Merge Thread #0 is trying to commit the merge and needs to acquire IndexWriter lock to do so, but the ArticleIngestor thread will not prevent it from acquiring the lock (since it releases the lock while inside wait()). Though I'm not sure why this thread is blocked (unless there are other threads that hold the lock). Do you have a small test case where the deadlock always happens? Mike Jeremy Volkman <jvolk...@gmail.com> wrote: > Just ran into this. I'm using Lucene 2.4 in the following manner: > > 1. Open IndexWriter > 2. Add documents > 3. Delete documents > 4. Close IndexWriter > > I haven't touched the out-of-the-box settings WRT merging. A JVM stacktrace > shows the following: > > "Lucene Merge Thread #0" daemon prio=10 tid=0x0000000050cb0c00 nid=0x44ce > waiting for monitor entry [0x0000000047a90000..0x0000000047a90b90] > java.lang.Thread.State: BLOCKED (on object monitor) > at > org.apache.lucene.index.IndexWriter.commitMerge(IndexWriter.java:3751) > - waiting to lock <0x00002aab25254cc8> (a > org.apache.lucene.index.IndexWriter) > at > org.apache.lucene.index.IndexWriter.mergeMiddle(IndexWriter.java:4240) > at org.apache.lucene.index.IndexWriter.merge(IndexWriter.java:3877) > at > org.apache.lucene.index.ConcurrentMergeScheduler.doMerge(ConcurrentMergeScheduler.java:205) > at > org.apache.lucene.index.ConcurrentMergeScheduler$MergeThread.run(ConcurrentMergeScheduler.java:260) > > "Indexing thread [ArticleIngestor]" prio=10 tid=0x000000004fd8d800 > nid=0x320f waiting on condition [0x0000000042d43000..0x0000000042d43c10] > java.lang.Thread.State: RUNNABLE > at org.apache.lucene.index.IndexWriter.doWait(IndexWriter.java:4491) > - locked <0x00002aab25254cc8> (a org.apache.lucene.index.IndexWriter) > at > org.apache.lucene.index.IndexWriter.finishMerges(IndexWriter.java:2790) > - locked <0x00002aab25254cc8> (a org.apache.lucene.index.IndexWriter) > at > org.apache.lucene.index.IndexWriter.closeInternal(IndexWriter.java:1632) > at org.apache.lucene.index.IndexWriter.close(IndexWriter.java:1588) > at org.apache.lucene.index.IndexWriter.close(IndexWriter.java:1562) > > finishMerges is a synchronized method, and commitMerge is also a > synchronized method. finishMerges is waiting for the "runningMerges" > collection to empty out before returning and subsequently giving up the > monitor it holds on the IndexWriter object. That collection won't ever > empty, however, because the merge thread is waiting for the IndexWriter > monitor. > > Am I missing something here? Could this be user error? > > Thanks, > Jeremy > --------------------------------------------------------------------- To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org For additional commands, e-mail: java-user-h...@lucene.apache.org
