Re: Does this perf output seem 'normal'?

On 28.04.2016 1:11, Peter Devoy wrote:
> I have now done a recording for 60 seconds during a batch of 1000
> requests and posted the results on a new issue on the Mapnik repo.
>
> Although Postgres still comes out on top in the perf results I
> struggle to believe this is a Postgres issue. But, if anyone is
> curious, the issue is here:
> https://github.com/mapnik/mapnik/issues/3414
>
> I may come back here if I don't have luck with the Mapnik developers.
>
> All the best
>
>
> Peter
> 3XE
> P: 01326 567155
> M: 07770 693662
> A: 3XE Ltd
> Tremough Innovation Centre
> PENRYN
> TR10 9TA
> 3XE Ltd · Registered in England and Wales · 9356871
>
>
> On 27 April 2016 at 16:45, Peter Devoy <peter(at)3xe(dot)co(dot)uk> wrote:
>>> If you really want to profile this, you should fire it off in a tight loop, using wget or ab2 or curl.
>> Thanks Jeff, that sounds like a smart idea. I will try later when I
>> have access to the server.
>>
>>> Hi! What do you want to see in perf stats? Maybe you can explain your problem more in details?
>> Hi Alex ) I am hoping to find out which function calls are taking the
>> longest because to the code is taking too long to only parse a small
>> XML file and do a few quick database queries. The last version of
>> this software was able to do a lot more in the same period of time. I
>> realise this most likely a client software issue but was hoping it may
>> be obvious whether Mapnik is spending too much time with Postgres.
>>
>> Unfortunately I may just have to try again to build Mapnik from source
>> so that I can get debugging/profiling output, etc. -- I was hoping to
>> avoid it because it is quite painful.
Can you expand node from perf report where 33.55% and 13.37% and 38.04%
38.04%.
PS Also do you know that in perf report second column is the total cpu
time share and not the first ? And that is why total sum of second
column is 100% but sum from first can be != 100%? First column is the %
of time when child function(procedure) is working relative to main proc
time.

--
Alex Ignatov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company