Does anyone know what the size of the queue in the server queue means?
By the definition of doc:
9.2.5. hbase.regionserver.compactionQueueSize The size of the compaction queue. This is the number of stores in the region that were destined for compaction.
Is the Store number (or save files? I heard two versions) of the register server needs to be compressed.
I have a job writing data in hotspot style using a serial key (not distributed). and I saw inside the metric history, finding that the compression-queue-size = 4 occurred at that time. This is theoretically impossible, since I have only one record store (serial key) at any time.
Then I go into the log, I find that there is some hint at the queue size> 0: Each large seal says "This choice was in the queue for 0sec
013-11-26 12: 28: 00 778 INFO [regionserver60020-smallCompactions-1385440028938] regionserver.HStore: Completed a large compaction of 3 files in f1 myTable.key.md5 .... in md5 .... ( size = 607.8 M), the total size for the storage is 645.8 M. This choice was in the queue for 0 seconds and took 39 seconds to complete.
Even more confusing is: doesn't multithreading support in an earlier version and just allocate each compaction job to a thread, for this reason there is a compaction queue?
Too bad there is no detailed explanation in the hbase doc.