Re: Strange error (Socket command option unknown) - Mailing list pgsql-general

From Carlos Moreno
Subject Re: Strange error (Socket command option unknown)
Date
Msg-id 3E5A9D2E.4070808@mochima.com
Whole thread Raw
In response to Strange error (Socket command option unknown)  (Carlos Moreno <moreno@mochima.com>)
Responses Re: Strange error (Socket command option unknown)  (Tom Lane <tgl@sss.pgh.pa.us>)
Re: Strange error (Socket command option unknown)  (Steve Crawford <scrawford@pinpointresearch.com>)
List pgsql-general
Tom Lane wrote:

>
>You should look very hard at the part of your code that deals with
>error reports from the server.
>
>It might be worth logging the whole session between client and
>server with tcpdump --- that'd be tedious to wade through, but
>it might give some evidence of what's happening.  It seems pretty
>clear that the client is sending an incomplete message, but knowing
>only one character of the broken message isn't much to help you
>debug it.  With a tcpdump trace you could see the whole packet ...
>

Ok, it would look like I am going to have to face the
terror of doing this!  :-)

However, I have zero experience on this; if I may,
I'd like to ask for some pointers on this tcpdump;

checking the man pages, I came up with:

tcpdump -i local port 5432

I then connected using psql -h localhost, as a test;
it seems to be showing only timing and length information.
What could I do?

The other thing is that this is only a quick test to first
see that I can figure out the information that tcpdump
gives me...  But the connection is not being done through
TCP/IP, since the application and postmaster are running
on the same physical machine.

So, how can I listen on local (i.e., AF_UNIX) sockets?
(which I assume is the way my application and the postgres
server communicate)

Thanks!

Carlos
--



pgsql-general by date:

Previous
From: "mike McGire"
Date:
Subject: Re: A few questions to real pgsql gurus
Next
From: Joe Tomcat
Date:
Subject: Re: 7.4?