[
https://issues.apache.org/jira/browse/HBASE-25082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17211235#comment-17211235
]
Bharath Vissapragada commented on HBASE-25082:
----------------------------------------------
[~reidchan] Sorry I was out for a bit, catching up now.
bq. for the append size why don't we use SizeHistogram but Counter.
Not sure I understand your approach. Histogram only captures the distribution
right? How does it give per-table information (which is the ask here). We'd
like to know, for each table how many appends happened and whats the cumulative
size of those appends. Let me know if I'm missing something.
> Per table WAL metrics: appendCount and appendSize
> -------------------------------------------------
>
> Key: HBASE-25082
> URL: https://issues.apache.org/jira/browse/HBASE-25082
> Project: HBase
> Issue Type: Improvement
> Components: metrics
> Affects Versions: 3.0.0-alpha-1, 2.3.3, 1.7.0, 2.4.0
> Reporter: Bharath Vissapragada
> Assignee: Bharath Vissapragada
> Priority: Major
> Labels: observability
> Fix For: 3.0.0-alpha-1, 2.3.3, 1.7.0, 2.4.0
>
>
> We were trying to drill down into WAL events at a table scope and figured
> that appendCount and appendSize are not exposed at a table scope. This would
> be a helpful metric.
> I ran the WAL benchmark tool with 10 threads and didn't see any measurable
> performance difference with the attached patch.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
