[
https://issues.apache.org/jira/browse/HBASE-14969?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15202854#comment-15202854
]
Gary Helmling commented on HBASE-14969:
---------------------------------------
[~carp84], the part I'm looking at is in {{CompactSplitThread}}, where the
{{CompactionThroughputControllerFactory}} used changed:
{code}
-import
org.apache.hadoop.hbase.regionserver.compactions.CompactionThroughputController;
-import
org.apache.hadoop.hbase.regionserver.compactions.CompactionThroughputControllerFactory;
+import
org.apache.hadoop.hbase.regionserver.throttle.CompactionThroughputControllerFactory;
+import org.apache.hadoop.hbase.regionserver.throttle.ThroughputController;
{code}
The snippet you pulled is from
o.a.h.h.regionserver.compactions.CompactionThroughputController.
I don't know why we have two {{CompactionThroughputControllerFactory}} classes,
but the o.a.h.h.regionserver.throttle one defaults to
{{PressureAwareCompactionThroughputController}} while the
o.a.h.h.regionserver.compactions one (previously used) defaults to
{{NoLimitCompactionThroughputController}}.
If the change wasn't intentional, maybe we should do a small follow up issue to
change the default back. Even better if we remove the duplicate class and just
use one.
> Add throughput controller for flush
> -----------------------------------
>
> Key: HBASE-14969
> URL: https://issues.apache.org/jira/browse/HBASE-14969
> Project: HBase
> Issue Type: Improvement
> Components: regionserver
> Affects Versions: 2.0.0, 1.3.0
> Reporter: Yu Li
> Assignee: Yu Li
> Fix For: 2.0.0, 1.3.0
>
> Attachments: HBASE-14969-addendum-branch-1.patch,
> HBASE-14969.branch-1.patch, HBASE-14969.patch, HBASE-14969_v10.patch,
> HBASE-14969_v2.patch, HBASE-14969_v3.patch, HBASE-14969_v4.patch,
> HBASE-14969_v5.patch, HBASE-14969_v6.patch, HBASE-14969_v9.patch,
> fd78628_9909808_compat_report.html, load-nothrottling.log, load-throttling.log
>
>
> In HBASE-8329 we added a throughput controller for compaction, to avoid spike
> caused by huge IO pressure like network/disk overflow. However, even with
> this control on, we are still observing disk utils near 100%, and by analysis
> we think this is caused by flush, especially when we increase the setting of
> {{hbase.hstore.flusher.count}}
> In this JIRA, we propose to add throughput control feature for flush, as a
> supplement of HBASE-8329 to better control IO pressure.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
