[rabbitmq-discuss] Statistics gathering

[Matthias Radestock]

Flavio,

have you been able to answer the questions below yourself, after further 
investigation?

Regards,

Matthias

On 25/07/13 13:18, Flavio Pompermaier wrote:
> So, actually I have just to tune sample_retention_policies to keep and
> aggregate statistics as long as I need!
> Is there a way to reset those statistics (during development I think I
> will make several tests in a day, but I think I'll keep stats for a day
> so I need to reset those couters..)?
> Moreover, I saw that detailed stats are available only for 10
> seconds..do you think it could be feasible to keep detailed data for an
> entire day or does it occupy too much space on disk?
> For example, I'd like to do something like:
>
> {rabbitmq_management, [{sample_retention_policies,
>                            [{global,   [{60, 5}, {3600, 60}, {86400,
> 1200}]},
>                             {basic,    [{60, 5},{3600, 60}, {86400, 1200}]},
>                             {detailed, [{10, 5},{3600, 60}, {86400, 1200}]}
>                          ]},
>
> What is not clear to me is which statistics provides each category..for
> example, detailed what information does affects?
> If i ask for /api/queues, will I receive more or less data if I had also
> detailed infos?
> Resolution defined in sample_retention_policies have different
> granularities in order to perform cumulative stats faster and save disk,
> right?
> Because I could configure {basic, [{{86400, 5}]} but then I keep all
> 5-secs samples for all the day and also compute aggregate stats will
> require to pass all the samples..(correct me if I'm wrong).
> Will HTTP api returns cumulative stats only? sample_age and sample_incr
> parameters are useful only I want all samples, right?
>
> Thanks for the support,
> Flavio