hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Enis Soztutar (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-15377) Per-RS Get metric is time based, per-region metric is size-based
Date Wed, 09 Mar 2016 02:11:40 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15377?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15186333#comment-15186333
] 

Enis Soztutar commented on HBASE-15377:
---------------------------------------

Thanks [~chenheng] for taking on this patch. It looks good, except that I think we can keep
"Get" (as opposed to GetTime) as the name of the metric for region server and region level,
and introduce "GetSize" as the new metric. I know that for scans we did "ScanSize" and "ScanTime",
but in this case, we can keep compatibility for the regionserver-level metric name. wdyt?
 

> Per-RS Get metric is time based, per-region metric is size-based
> ----------------------------------------------------------------
>
>                 Key: HBASE-15377
>                 URL: https://issues.apache.org/jira/browse/HBASE-15377
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Enis Soztutar
>            Assignee: Heng Chen
>             Fix For: 2.0.0, 1.3.0, 1.4.0
>
>         Attachments: HBASE-15377.patch, HBASE-15377_v1.patch
>
>
> We have metrics for Get operations at the region server level and region level. 
> {code}
>    "Get_num_ops" : 4837505,
>     "Get_min" : 0,
>     "Get_max" : 296,
>     "Get_mean" : 0.2934618155433431,
>     "Get_median" : 0.0,
>     "Get_75th_percentile" : 0.0,
>     "Get_95th_percentile" : 1.0,
>     "Get_99th_percentile" : 1.0,
> {code}
> and 
> {code}
>    "Namespace_hbase_table_meta_region_1588230740_metric_get_num_ops" : 103,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_min" : 450,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_max" : 470,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_mean" : 450.19417475728153,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_median" : 460.0,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_75th_percentile" : 470.0,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_95th_percentile" : 470.0,
>     "Namespace_hbase_table_meta_region_1588230740_metric_get_99th_percentile" : 470.0,
> {code}
> The problem is that the report values for the region server shows the latency, versus
the reported values for the region shows the response sizes. There is no way of telling this
without reading the source code. 
> I think we should deprecate response size histograms in favor of latency histograms.

> See also HBASE-15376. 



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

Mime
View raw message