[Radiant] Trouble deploying Radiant on a shared host

Matt Bauman mbauman at gmail.com
Sun Apr 29 15:59:20 CDT 2007


Hello all,

I'm trying to deploy Radiant on Site5's shared hosting, but I can't  
seem to get it to go.  I was wondering if anyone had any pointers on  
how to try to debug this.  Here's how I have it setup:

I created a gems folder in my home, declared it my GEM_HOME in  
my .bash_profile and gemhome in .gemrc (as well as prepending it to  
my gempath in both places), and installed radiant without any  
troubles to it.  I unpacked radiant (./gems/bin/radiant --database  
mysql radiant).  I setup database.yml and pre-populated it with the  
bootstrap script.  All is gravy.  A very quick test of ./script/ 
server shows that it's running without any trouble at all.  The  
server runs Ruby 1.8.4.

But: I can't get the cgi scripts working.  I get an Apache error  
500.  Fast CGI repeatedly exits (with status 1) until it gets its  
restart interval rolled back to 600 seconds, and then it times out  
with the error:  FastCGI: comm with (dynamic) server "/home/mbaumann/ 
public_html/test/dispatch.fcgi" aborted: (first read) idle timeout  
(30 sec).  And plain-jane cgi gives up the chase immediately with the  
error:  Premature end of script headers: /home/mbaumann/public_html/ 
radiant/dispatch.cgi. The scripts themselves don't throw errors when  
run via the command line (./public/dispatch.(f)cgi) -- they simply  
exit with status 0.

As a quick test, I quickly made a rails app ("rails foo") and made a  
simple controller (rails -v yields version 1.2.3).  That worked, with  
both fcgi and cgi (and fcgi was indeed faster).  So I tried replacing  
the public folder in the radiant install with the one from "foo".   
Same error.  I tried freezing radiant (from the gem) and replacing  
the vendor/radiant/vendor/rails with the frozen counterpart from the  
working "foo".  Same error.  Permissions seem to be set correctly (I  
have a umask of 0022 and the cgi scripts run under my user), but I  
tried chmod'ing the entire install to 755.  Same error.  I tried  
mucking around to my wits end in the public folder, reinstalling,  
retesting, reinstalling... same error.  I've looked around at all the  
guides to deploying on shared hosts, trying all their debugging tips,  
and I get the same error.  I've searched through the mailing list  
archives, but nothing there, either.

I'll file a ticket with Site5, too, but to be honest, from my (naive)  
perspective, it really looks like it's a problem on Radiant's side of  
things.  I've posted on the discussion board to no avail , but there  
are other users there running the mental release candidates without  
troubles.  Is the Mental RC1 revision still available?  I can test  
freezing to that revision (if I get the number) since I know that's  
worked for another Site5 user.

Sorry for the terribly long email... but I'm really flummoxed at this  
point.  Any ideas on how to further debug this?  Or ways of  
convincing Site5 that it's really their problem?
Thanks!
Matt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.radiantcms.org/pipermail/radiant/attachments/20070429/9ddb7484/attachment.html


More information about the Radiant mailing list