You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Otis Gospodnetic (Updated) (JIRA)" <ji...@apache.org> on 2012/02/07 04:55:02 UTC
[jira] [Updated] (HBASE-5301) Some RegionServer metrics have really
confusing names
[ https://issues.apache.org/jira/browse/HBASE-5301?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Otis Gospodnetic updated HBASE-5301:
------------------------------------
Component/s: metrics
Description:
Mikael Sitruk commented on this back in Nov 2011 and after looking at this I completely agree with him. For example, "flushSize_avg_time" makes no sense. "flushSize" is in bytes, so is this the average flush size? Or the average time per flush? In which case, why not call the measure "flush_avg_time". But to add to the confusion there is already a "flushTime_avg_time" metric. There is also "flushTime_num_ops" and "flushSize_num_ops" that are confusing. Is the former the number of flushes? In which case, why have "time" in the metric name?
On 11/22/11 5:23 PM, "Mikael Sitruk" <mi...@gmail.com> wrote:
Hi
I have enabled metrics on Hbase cluster (0.90.1), and mapped the metrics
to
3 categories (missing, Present but not documented/Incomplete
documentation,Ok) according to their status in the book (
http://hbase.apache.org/book.html#hbase_metrics). Is it possible to udpate
the book accordingly?
It seems also that rpc metrics are not documented at all.
And now some questions on the metrics:
I can see some metrics present a num_ops and avg_time suffix (like rpc)
but
it seems that for certain metrics is it totally unclear (to me at least)
or
their name is missleading - for example what
means compactionTime_avg_time/compactionTime_num_ops? or
flushSize_avg_time
and flushSize_num_ops? I mean I would have understood compaction_avg_time
and flushSize or flush_avg_time.
was:
Mikael Sitruk commented on this back in Nov 2011 and after looking at this I completely agree with him. For example, "flushSize_avg_time" makes no sense. "flushSize" is in bytes, so is this the average flush size? Or the average time per flush? In which case, why not call the measure "flush_avg_time". But to add to the confusion there is already a "flushTime_avg_time" metric. There is also "flushTime_num_ops" and "flushSize_num_ops" that are confusing. Is the former the number of flushes? In which case, why have "time" in the metric name?
On 11/22/11 5:23 PM, "Mikael Sitruk" <mi...@gmail.com> wrote:
Hi
I have enabled metrics on Hbase cluster (0.90.1), and mapped the metrics
to
3 categories (missing, Present but not documented/Incomplete
documentation,Ok) according to their status in the book (
http://hbase.apache.org/book.html#hbase_metrics). Is it possible to udpate
the book accordingly?
It seems also that rpc metrics are not documented at all.
And now some questions on the metrics:
I can see some metrics present a num_ops and avg_time suffix (like rpc)
but
it seems that for certain metrics is it totally unclear (to me at least)
or
their name is missleading - for example what
means compactionTime_avg_time/compactionTime_num_ops? or
flushSize_avg_time
and flushSize_num_ops? I mean I would have understood compaction_avg_time
and flushSize or flush_avg_time.
Labels: metrics (was: )
> Some RegionServer metrics have really confusing names
> -----------------------------------------------------
>
> Key: HBASE-5301
> URL: https://issues.apache.org/jira/browse/HBASE-5301
> Project: HBase
> Issue Type: Bug
> Components: metrics
> Reporter: Doug Meil
> Labels: metrics
>
> Mikael Sitruk commented on this back in Nov 2011 and after looking at this I completely agree with him. For example, "flushSize_avg_time" makes no sense. "flushSize" is in bytes, so is this the average flush size? Or the average time per flush? In which case, why not call the measure "flush_avg_time". But to add to the confusion there is already a "flushTime_avg_time" metric. There is also "flushTime_num_ops" and "flushSize_num_ops" that are confusing. Is the former the number of flushes? In which case, why have "time" in the metric name?
> On 11/22/11 5:23 PM, "Mikael Sitruk" <mi...@gmail.com> wrote:
> Hi
> I have enabled metrics on Hbase cluster (0.90.1), and mapped the metrics
> to
> 3 categories (missing, Present but not documented/Incomplete
> documentation,Ok) according to their status in the book (
> http://hbase.apache.org/book.html#hbase_metrics). Is it possible to udpate
> the book accordingly?
> It seems also that rpc metrics are not documented at all.
> And now some questions on the metrics:
> I can see some metrics present a num_ops and avg_time suffix (like rpc)
> but
> it seems that for certain metrics is it totally unclear (to me at least)
> or
> their name is missleading - for example what
> means compactionTime_avg_time/compactionTime_num_ops? or
> flushSize_avg_time
> and flushSize_num_ops? I mean I would have understood compaction_avg_time
> and flushSize or flush_avg_time.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira