On 11/07/2014 12:38 PM, Dieter Plaetinck wrote:
On Fri, 07 Nov 2014 09:56:47 -0800
Rob Miller <rmil...@mozilla.com> wrote:
> On 11/07/2014 09:51 AM, Dieter Plaetinck wrote:
>> anyway, i just did some crude testing.
>> on an intel xeon E5-2690 @ 2.90GHz heka consumes 70~80% cpu typically, when
i send around 75k timers/s and 40k counters/s and once in a while it'll drop some
packets from the udp buffer. (like 10k every minute or so)
>> this is with stat accumulator every 60s; enabling dashboard and looking at a
realtime graph with 3 series tracked does not seem to impact cpu at all.
>>
>> again, quick and crude test. but it seems pretty good and at least in the
same order of what i got with statsdaemon.
>>
> Interesting. Did you have maxprocs set, or was it running on a single core?
I'd be curious to know what you see with maxprocs values of 1, 2, 3, and 4.
this was with default, 1.
i'm not gonna do much more testing though, i'm gonna deploy heka's and
statsdaemons to the servers themselves to spread the load, so i don't need high
perf anymore.
I strongly suggest you try again at least once w/ maxprocs set to >1. When
you're on a single core, then the UDP network activity and the stat aggregation
are competing for the use of that single core. If you make more than one core
available, then one of them can handle the network activity and another can handle
the stat aggregation, you will likely see a very significant increase in
throughput.
-r
_______________________________________________
Heka mailing list
Heka@mozilla.org
https://mail.mozilla.org/listinfo/heka
