Problems with FastCGI in 4.4-beta1

Jean-Francois Doyon Jean-Francois.Doyon at CCRS.NRCAN.GC.CA
Thu Oct 28 16:05:21 EDT 2004


Frank,

Well, first of all, I'm actually mostly interested in the regular CGI mode,
so that's a buzz kill for me :)

I'll keep it in mind though, might still be nice for WMS, though in our case
the CGI version gets far heavier use than our WMS.

My test was calling mapserv without any parameters, just for testing,
especting the standard "no query set" error message.

The FastCGI compiled version works just fine in standard CGI mode.

I too was rather intrigued by the fact that the process seemed to keep dying
unexpectedly, something rather hard to track down.

Although, now that you mention the WMS thing, could it be that the simple
fact that I was using non WMS requests was causing problem.

Under your current implementation, what is the expected behavior if a
non-WMS request comes in through FastCGI ?

I'll try a more complex configuration along the lines of what Paul suggested
in the mean time.

Honestly I don't NEED it, but I'm always glad to get the extra performance
anywhere I can, and this should normally be pretty straightforward to setup
:P

thanks,
J.F.


-----Original Message-----
From: Frank Warmerdam [mailto:warmerdam at pobox.com]
Sent: October 28, 2004 3:34 PM
To: Jean-Francois.Doyon at CCRS.NRCAN.GC.CA
Cc: MAPSERVER-USERS at LISTS.UMN.EDU
Subject: Re: [UMN_MAPSERVER-USERS] Problems with FastCGI in 4.4-beta1


Jean-Francois Doyon wrote:
> Hello,
>
> I'm trying to get the FactCGI version working, but am not having any luck:
>
> [Thu Oct 28 14:45:40 2004] [notice] Apache/1.3.31 (Unix) mod_fastcgi/2.4.2
> configured -- resuming normal operations
> [Thu Oct 28 14:45:40 2004] [notice] Accept mutex: sysvsem (Default:
sysvsem)
> [Thu Oct 28 14:45:44 2004] [warn] FastCGI: server
> "/usr/local/apache/cgi-bin/mapserv" (pid 9471) terminated by calling exit
> with status '1'

...

> Where the mapserv executable was in apache's default cgi-bin directory.
>
> I'm on CentOS (RHEL 3), with apache 1.3.31 and FastCGI 2.4.2 ...
>
> Anybody else tried this yet? Any success?

JF,

I am locally working with it on FC2 with Apache2.  It looks like mapserv is
terminating with an exit(1) somewhere somewhat early.   Does the mapserv
work properly in plain cgi-mode?  The same executable build should operate
in
either mode. I usually like to start by debugging in plain cgi mode to
verify
that works, and then invoke it in fastcgi mode.

Also, are you using it in WMS or "traditional mapserver" mode?  Currently
only
WMS is formally supported.  I do hope to fix up traditional mode though.

 > I'm starting to look at and evaluate 4.4 as an upgrade from 3.6.6 (Yup,
I'm
 > still using it, works so well!), and I just have a couple of questions
...
 >
 > 1) Is FastCGI deemed stable and reliable in 4.4 ? I'd LOVE to use it.

I think we would have to consider it an experimental feature in 4.4 since it
is so new.  But I am keen to fix any problems.  I'd love you to use it, and
point out problems!

BTW, are you interested in FastCGI primarily for database connection
preservation or what?  I think that is the main case where it offers a
substantial performance boost.

 > 2) If I use FastCGI, I need it thread safe enough?

FastCGI with the main mapserver executable is effectively single threaded.
Each fastcgi process accepts one request at a time, and finishes it before
accepting another.  Several fastcgi processes can run at the same time of
course.

Best regards,

--
---------------------------------------+------------------------------------
--
I set the clouds in motion - turn up   | Frank Warmerdam,
warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent



More information about the mapserver-users mailing list