The FastCGI gateway on my server went down earlier today, and as a result the ICHC app stopped working for a while. I moved the feed scraper to a separate user account, so it should be less likely to go down now.
Nginx uses a fixed number of FastCGI sockets (currently 5) for each user running on the server. Certain processes (one of which seems to be the PrMac importer I use at MacMegasite) will cause it to run out of FastCGI connections resulting in a ‘Bad Gateway’ error. By moving the app feed to a separate user, it now has its own set of FastCGI sockets so other processes should be less likely to bring it down.
If you do ‘ps ax’ on a DreamHost VPS shell account, you can see how many FastCGI processes are running:
3927 ? Ss 0:00 /dh/cgi-system/php5.cgi -b /home/sampo/.php.sock 3932 ? S 0:00 /dh/cgi-system/php5.cgi -b /home/jimi/.php.sock 3933 ? S 0:00 /dh/cgi-system/php5.cgi -b /home/newapp/.php.sock 3934 ? S 0:00 /dh/cgi-system/php5.cgi -b /home/newapp/.php.sock 3935 ? S 0:01 /dh/cgi-system/php5.cgi -b /home/mysites/.php.sock 3936 ? S 0:00 /dh/cgi-system/php5.cgi -b /home/newapp/.php.sock 3937 ? S 0:02 /dh/cgi-system/php5.cgi -b /home/mysites/.php.sock 3938 ? S 0:00 /dh/cgi-system/php5.cgi -b /home/newapp/.php.sock