Re: making sense of the timer(s) sequence

On Wed, Aug 20, 2008 at 10:04 PM, Aleksandar Lazic <al-haproxy#none.at> wrote:
> Hi Jeffrey,
>
> On Die 19.08.2008 17:36, Jeffrey 'jf' Lim wrote:
>>
>> hi guys, i'm hoping to get a better sense of the timers and values in
>> the logs, and am looking to clarify my own understanding in terms of
>> both the sequence, and the meaning of the various timers. The following
>> is what i've arrived at. Is the following understanding correct?
>>
>> In terms of a timeline, this is what happens: (assuming option
>> httpclose, to make things simpler)
>>
>> half of Tw - a connection comes in, and may spend some time waiting in
>> the listener's queue (*INCLUDING connection establishment)
>> Tq - client sends request over
>> the other half of Tw - time in the server queue
>> Tc - syn, syn-ack delay to server
>> Tr - server header response time
>> Td (which is not logged directly)
>>
>> Am i correct?
>
> http://haproxy.1wt.eu/download/1.3/doc/haproxy-en.txt
> => 4.2.3) Timing events
>
> Please check it by your own ;-)
> ###
> <snip>

I've read this already. Otherwise, how did i arrive at the definitions of the various 'T's? I am specifically looking at the sequence here, and various idiosyncracies that are not exactly apparent in the documentation - for eg. the fact that 'Tw' is split up into 2 (see my timeline) - *and* the fact that 'Tw' seems to include the client syn, syn-ack while *not* including the server syn, syn-ack. I'd basically like to get clarifications for these 2 points, if that makes my question clearer and more specific.

thanks,
-jf Received on 2008/08/21 04:22