[
https://issues.apache.org/jira/browse/HBASE-16698?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15526894#comment-15526894
]
Yu Li commented on HBASE-16698:
-------------------------------
I don't have much benchmarking number at hand, but from our online cluster we
see no regression on put average time.
Notice that the problem exists even if there's only one region but many
parallel writes (yes, after a relook I think I stated something wrong, the
issue stands even if there's only one region), allow me to quote the existing
code flow on append handling:
{noformat}
RingBufferEventHandler grab one append
-> FSHLog#append is called
-> FSWALEntry#stampRegionSequenceId is called
-> One CountDownLatch is released
-> RingBufferEventHandler grab another append
-> Another CountDownLatch is released
-> Repeat
{noformat}
so the main problem is *sequential* appends and the logic that getting MVCC has
to wait for the relative append to finish.
I'll supply some perf number with YCSB, but should be days later because of
some headache online issues... Or it will be highly appreciated if anyone could
offer some help on the bench testing.
> 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: 1.1.6, 1.2.3
> Reporter: Yu Li
> Assignee: Yu Li
> Attachments: 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)
