503 Service Unavailable
Just ran a default install, and when trying to access the admin pages or run the test PHP script, it throws up an "Error 503 Service Unavailable," and the errorlog says:
2003-08-04 12:12:16.167 [WARN] Start FCGI [fcgiAdminPHP]: failed to bind to localhost:7202, fast cgi might be running,
rebuild php binary
Can you please tell me more about your system? Which distrubtion? Redhat? Suse?
We were informed that our pre-built php binary does not work well on Redhat 9 because of newer version of glibc. So please rebuild php binary from source if you can.
Please use PHP 4.3.1, 4.3.2 is buggy. build it as following:
./configure --enable-fastcgi --with-config-file-path=../conf
cp -f sapi/cgi/php [lsws_home]/admin/fast-cgi/admin_php
cp -f sapi/cgi/php [lsws_home]/fast-cgi/php
Thanks. I have Redhat 8, but have updated glibc since I first installed so I guess therein lies the problem.
Downloaded 4.3.1, but I get this error now during configure:
checking for gcc... no
checking for cc... no
configure: error: no acceptable cc found in $PATH
You are welcome.
Did you install gcc?
Did you set PATH env correctly?
With the new glibc, you may also have problem during compilation. If it complains about "errno" in MySQL library, just modify
add a line at the beginning
We just fixed a bug in PHP binary related to the new glibc library and it runs properly on redhat 9 now.
You can download our repackaged binary package and try again. See if it is fixed.
Okay thanks for your help so far. Managed to build PHP as you instructed, still getting a 503 though. Error log throws up:
2003-08-04 20:50:40.600 [WARN] Start FCGI [fcgiAdminPHP]: failed to bind to localhost:7202, fast cgi might be running,
change the port for fcgiAdminPHP
Two possible causes:
1. the old admin_php is still running:
2. Port 7202 is use by other daemon process.
Please post your result.
This is just wierd and frustrating now...
Wasn't possibility number 1.
netstat -an | grep 7202 revealed it was being used, so I changed it to 5158 (which wasn't being used)
netstat -an | grep 5158 now reveals:
tcp 0 0 127.0.0.1:5158 0.0.0.0:* LISTEN
So its almost as if it is listening on this port then it tries to start listening (again?!) when a request is put through??
Need to test your PHP binary.
Do not give up yet. :-)
What you observed is normal. Once you try to access the admin interface, the web server will bind (use) port 5158 and start PHP, so when you do netstat again, it shows port 5158 is in use, if you kill all the running PHP processes, that port will be released.
For some reason the Fast CGI interface on PHP side is broken and no request can be completed, when that happens, web server will try to restart PHP engine and try again, but the port is still used by left over PHP processes. That is the reason why you saw that warning.
First we need to make sure your PHP binary is ok. Try it form command line first, see if there is any error messages, run "php -i" to make sure Fast CGI interface is enabled (configured with --enable-fastcgi).
If configuration file path for PHP is set to /usr/local/lib (it is the default value, my instrution change it to ../conf), make sure the notice logging is not enabled, it will break the fast CGI interface.
Then turn on server's debug logging by editing file "lsws/conf/httpd_config.xml", change "debugLevel" to 10. stop server, kill all PHP processes, start server again and access the PHP test page (not the admin interface).
If you still get 503, compress the error log and send it to email@example.com.
We will try our best to help you.
Thank you for your patience.
Still not coming up trumps, I have emailed you the errorlogs.
|All times are GMT -7. The time now is 06:06 PM.|