perl-modperl mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Michael Koob" <>
Subject RE: Intermittent Error - Setting Cookies on POST request redirect response
Date Thu, 08 May 2003 19:08:02 GMT

> -----Original Message-----
> From: James G Smith [mailto:JGSmith@TAMU.Edu] 
> Sent: Thursday, May 08, 2003 2:46 PM
> To: Michael Koob
> Cc: 'mod_perl Mailing List'
> Subject: Re: Intermittent Error - Setting Cookies on POST 
> request redirect response 
> This describes the problem we're seeing on a separate 
> (non-mod_perl, php actually) system.  IE 6 will log in to the 
> site and then when a customer goes to certain pages or 
> performs certain functions, it logs out or does other odd 
> things.  No other browsers are reported as being a problem, 
> as far as I can tell (I don't work customer support).  It 
> looks like a browser problem (one of the many in IE 6, apparently).

	I have noted the same issue as being present in at least IE 5.5
so I am not sure that my information will help you isolate
an offending browser. Other than IE in general of course :)

> Do you mind if I use parts of your e-mail in an internal 
> e-mail on this subject?  It points to some of the possible 
> causes and helps narrow the field of alternatives for us 
> (such as not setting a cookie on each request -- a little 
> less secure, but workable).
> -- 
> James Smith <JGSmith@TAMU.Edu>, 979-862-3725
> Texas A&M CIS Operating Systems Group, Unix

	I am putting together a test script using Perl/LWP to attempt to
recreate the problem. If it is possible to recreate it via that script
then I can
at least attempt some further debugging. And it might eliminate IE as
the problem in that case. You can feel free to use the information if it
helps you. I am thinking that my issue may actually be database latency.
I have noted that the time difference between requests from the initial
POST to the request for the redirected page is currently 1s or less,
which is extremely quick and it is often longer on successful requests.
This makes me think that the session may not be getting stored fast
enough for it to be found on the second request. If the database
transaction is held up at all on the first transaction then the second
could be invalid. This may be reaching a bit, but if I can produce the
error with the script it will be easy to debug the fact that the session
could not be found in the database. The debugging code is already in the
production server so if it is happening I will know eventually.



View raw message