On Mon, Feb 08, 2010 at 12:35:32PM -0800, Paul Hirose wrote:
> The error earlier reported during -dev7 about not compiling with
> DEBUG_FULL is still present in rc1 and ss-20100208.  As that msg
> indicated, changing it f rom req to rep allowed the compile.

grrr, it seems I lost the patch. Care to resend it please ?

> Aside from DEBUG_MEMORY, it looks like most of the other
> DEBUG_FSM/HASH/AUTH are all related to "mode http".  Since I'm
> strictly "mode tcp" I'm guessing only DEBUG_FULL and DEBUG_MEMORY are
> of use to me, although DEBUG_MEMORY only seems to activate a couple of
> signal catchers, so it's not a big thing for me right now.
> 
> The main reason I'm trying a DEBUG_FULL is I'm getting reports from my
> clients of not being able to connect, or being disconnected.  But my
> haproxy logs (option tcplog) don't show any sign of it.  Even if I had
> hit some limit (connections, whatever), wouldn't there be an entry in
> the haproxy logs - denying the incoming connection or whatever, but
> there'd be some record of it?

Not necessarily. DEBUG_* macros are not used to debug traffic but for
developers to more easily debug the integration of their code in haproxy.
There is no specification of what macro does what, and it can change with
time.

> I admit, being buried in the logs is a bit tough, so I'm not sure yet.
>  I don't see any obvious non "--" status entries in my log (which is
> kinda where I'm starting.)
> 
> Any hints on tracking this down would be appreciated :)  I realize
> it's a bit vague, but it's all I've got to go on right now.

If your users can't connect, you won't find any log since it means
the connection failed to establish before reaching haproxy. If they
get disconnected, that depends if they have seen traffic or not. If
they could get some data, you definitely have the logs, but you'll
probably need their IP and the date to find anything in the logs.

They may fail to connect if your global maxconn is too low for example.

If you get the chance to find a user who can reliably reproduce the
issue, do not hesitate, start a tcpdump capture and you'll be fixed !

Willy


Reply via email to