[
https://issues.apache.org/jira/browse/HBASE-16698?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15596488#comment-15596488
]
Enis Soztutar commented on HBASE-16698:
---------------------------------------
On master, when I do a jstack with some load, almost all the handlers are
waiting for sync(), and since the memstore insert does not happen until sync()
completes, we do not have to wait for the latch. Actually, the latch can be
removed as well (for Durability.SYNC_WAL). For async, we still have to have the
latch I think.
The other reason I was asking about this is that I have a hacked up patch which
divides the batchMutate() into 3 phases. The first phase is the parts to
compute the WALEdit to write, second part is the actual append + sync() and the
last part is all other steps after sync() (memstore insert, etc). This
architecture makes it possible to do the flush pipelining idea talked about
here: http://pandis.net/resources/vldb10aether.pdf (section 4). The theory is
that we can free up the handler to process more requests while we are waiting
for the sync to happen. After sync some other handler or thread will complete
the work.
> Performance issue: handlers stuck waiting for CountDownLatch inside
> WALKey#getWriteEntry under high writing workload
> --------------------------------------------------------------------------------------------------------------------
>
> Key: HBASE-16698
> URL: https://issues.apache.org/jira/browse/HBASE-16698
> Project: HBase
> Issue Type: Improvement
> Components: Performance
> Affects Versions: 1.2.3
> Reporter: Yu Li
> Assignee: Yu Li
> Fix For: 2.0.0
>
> Attachments: HBASE-16698.branch-1.patch,
> HBASE-16698.branch-1.v2.patch, HBASE-16698.branch-1.v2.patch,
> HBASE-16698.patch, HBASE-16698.v2.patch, hadoop0495.et2.jstack
>
>
> As titled, on our production environment we observed 98 out of 128 handlers
> get stuck waiting for the CountDownLatch {{seqNumAssignedLatch}} inside
> {{WALKey#getWriteEntry}} under a high writing workload.
> After digging into the problem, we found that the problem is mainly caused by
> advancing mvcc in the append logic. Below is some detailed analysis:
> Under current branch-1 code logic, all batch puts will call
> {{WALKey#getWriteEntry}} after appending edit to WAL, and
> {{seqNumAssignedLatch}} is only released when the relative append call is
> handled by RingBufferEventHandler (see {{FSWALEntry#stampRegionSequenceId}}).
> Because currently we're using a single event handler for the ringbuffer, the
> append calls are handled one by one (actually lot's of our current logic
> depending on this sequential dealing logic), and this becomes a bottleneck
> under high writing workload.
> The worst part is that by default we only use one WAL per RS, so appends on
> all regions are dealt with in sequential, which causes contention among
> different regions...
> To fix this, we could also take use of the "sequential appends" mechanism,
> that we could grab the WriteEntry before publishing append onto ringbuffer
> and use it as sequence id, only that we need to add a lock to make "grab
> WriteEntry" and "append edit" a transaction. This will still cause contention
> inside a region but could avoid contention between different regions. This
> solution is already verified in our online environment and proved to be
> effective.
> Notice that for master (2.0) branch since we already change the write
> pipeline to sync before writing memstore (HBASE-15158), this issue only
> exists for the ASYNC_WAL writes scenario.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
