Good point Gordon, I was thinking the # of open scanners/mutators would
serve that purpose, but as you point out I should really have another
counter for # of opened scanners/mutators since the last "dump stats"
command.

Also the master will store per server and per table stats in something like
RRDTool <http://oss.oetiker.ch/rrdtool/index.en.html>, so that should
provide a sliding window of historical data.

-Sanjit

On Mon, Apr 12, 2010 at 3:05 PM, Gordon <[email protected]> wrote:

> Looks good ... one point is that for any of the aggregate stats (rates,
> avg, min, max) it's always good to have a sample size or number of
> observations since a average (or equivalently a rate for binary data) based
> on a small number of points is very different than for larger samples. This
> is especially true for the process control view of the system that looks at
> trends in these stats over a window of several periods.
>
> On Mon, Apr 12, 2010 at 7:15 PM, Sanjit Jhala <[email protected]> wrote:
>
>> One of the features we intend to add to release 0.9.4.0 is a monitoring
>> system.
>> The general idea is that the master will periodically issue something like
>> a "dump stats" command to the RangeServers and collect the responses.
>> The data collected will be used to make load balancing/range
>> re-distribution decisions as well as exposed via the monitoring interface so
>> that admins can monitor the health of the Hypertable cluster.
>> The Master can rollup stats to a per-RangeServer and per-table level since
>> that will probably make the most sense from a health check point of view.
>>
>> I propose the following list of *per RangeServer stats:*
>> -#ranges
>> -cpu load
>> -memory usage
>> -disk stats
>> -network stats
>> -block cache size
>> -query cache size
>> -#open file handles
>> -block cache hit rate
>> -query cache hit rate
>>
>> as well as a set of *per Range stats*:
>> -name/range id
>> -qps , bytes read per sec
>> -writes/s , bytes written/s
>> -#open scanners
>> -(min, max, avg) scanner lifetimes
>> -#open mutators
>> -(min, max, avg) mutator lifetimes
>>
>> Any thoughts on other useful stats for the monitoring console, range
>> balancing or on the monitoring framework in general would be very welcome.
>>
>> -Sanjit
>>
>>  --
>> You received this message because you are subscribed to the Google Groups
>> "Hypertable Development" group.
>> To post to this group, send email to [email protected].
>> To unsubscribe from this group, send email to
>> [email protected]<hypertable-dev%[email protected]>
>> .
>> For more options, visit this group at
>> http://groups.google.com/group/hypertable-dev?hl=en.
>>
>
>
>
> --
> Gordon Rios -- Cork Constraint Computation Centre
> http://www.4c.ucc.ie/web/people.jsp?id=144
> http://www.linkedin.com/in/gordonrios
>
> --
> You received this message because you are subscribed to the Google Groups
> "Hypertable Development" group.
> To post to this group, send email to [email protected].
> To unsubscribe from this group, send email to
> [email protected]<hypertable-dev%[email protected]>
> .
> For more options, visit this group at
> http://groups.google.com/group/hypertable-dev?hl=en.
>

-- 
You received this message because you are subscribed to the Google Groups 
"Hypertable Development" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/hypertable-dev?hl=en.

Reply via email to