Hi Guys,
after long googling and trying to find the right setup for nginx and fastcgi
we came to three settings:
fastcgi_buffers, fastcgi_buffer_size and fastcgi_read_timeout;

According to HttpFcgiModule manual the buffer size is equal to the size
of one buffer in fastcgi_buffers by default. Are this settings correct because a few hours later we are still getting BadGateway error ?

We are using Feebsd 8.2-STABLE amd64 jails and nginx from ports
sysctl hw.pagesize
hw.pagesize: 4096

fastcgi_connect_timeout 60;
fastcgi_send_timeout 20;
fastcgi_read_timeout 60;
fastcgi_buffers 512 4k;
fastcgi_buffer_size 4k;
fastcgi_busy_buffers_size 256k;
fastcgi_temp_file_write_size 256k;
fastcgi_intercept_errors on;
fastcgi_ignore_client_abort off;

limit_req_zone $binary_remote_addr zone=antiddosphp:10m rate=1r/s;
limit_req_zone $binary_remote_addr zone=antiddos:10m rate=10r/s;
open_file_cache max=100000 inactive=40s;
open_file_cache_valid 60s;
open_file_cache_min_uses 2;
open_file_cache_errors on;
client_max_body_size 150m;

ignore_invalid_headers on;
recursive_error_pages on;
sendfile on;
tcp_nopush on;
tcp_nodelay on;

client_body_timeout 25;
client_header_timeout 5;
keepalive_timeout 25 5;
send_timeout 25;
reset_timedout_connection on;

Here is php-fpm settings
emergency_restart_threshold = 0
emergency_restart_interval = 0
process_control_timeout=0

pm=dynamic
pm.max_children=65 pm.start_servers=30 pm.min_spare_servers=30 pm.max_spare_servers=65 pm.max_requests=10000 request_slowlog_timeout = 15s