[jira] [Commented] (HBASE-16162) Unnecessary push active segments to pipeline

["Anoop Sam John (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-16162?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15358615#comment-15358615 ] 

Anoop Sam John commented on HBASE-16162:
----------------------------------------

It is not same Active getting pushed.  Because this push happens under a lock ie. the region level write lock. 
There is no extra size check in flushInMemory after getting this lock. So the previous initiated in memory flush was waiting and acquire lock and will push the new active to pipeline.  This might not be empty also.  Because the time gap btw the 1st in memory compactor thread release the lock and the 2nd one acquire it, there might be some write to memstore happened.   We will end up in flush of small sized active segments.   

> Unnecessary push active segments to pipeline
> --------------------------------------------
>
>                 Key: HBASE-16162
>                 URL: https://issues.apache.org/jira/browse/HBASE-16162
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Anoop Sam John
>            Assignee: Anoop Sam John
>
> We have flow like this
> {code}
> protected void checkActiveSize() {
>     if (shouldFlushInMemory()) {
>          InMemoryFlushRunnable runnable = new InMemoryFlushRunnable();
>       }
>       getPool().execute(runnable);
>     }
>   }
> private boolean shouldFlushInMemory() {
>     if(getActive().getSize() > inmemoryFlushSize) {
>       // size above flush threshold
>       return (allowCompaction.get() && !inMemoryFlushInProgress.get());
>     }
>     return false;
>   }
> void flushInMemory() throws IOException {
>     // Phase I: Update the pipeline
>     getRegionServices().blockUpdates();
>     try {
>       MutableSegment active = getActive();
>       pushActiveToPipeline(active);
>     } finally {
>       getRegionServices().unblockUpdates();
>     }
>     // Phase II: Compact the pipeline
>     try {
>       if (allowCompaction.get() && inMemoryFlushInProgress.compareAndSet(false, true)) {
>         // setting the inMemoryFlushInProgress flag again for the case this method is invoked
>         // directly (only in tests) in the common path setting from true to true is idempotent
>         // Speculative compaction execution, may be interrupted if flush is forced while
>         // compaction is in progress
>         compactor.startCompaction();
>       }
> {code}
> So every write of cell will produce the check checkActiveSize().   When we are at border of in mem flush,  many threads doing writes to this memstore can get this check to pass.  Yes the AtomicBoolean is still false only. It is turned ON after some time once the new thread is started run and it push the active to pipeline etc.
> The time gap btw the size check return true and the block updates.   In btw all updates to memstore would have kicked this in memory flush
> So all these threads can push the same active to pipeline.  We unnecessarily passing through same segment many times. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)