I wonder in what environment Tailwatchd doing restarts.

The problem is that we have monitors for Nginx service which (Nginx) opens ~40k files.

If we restart from the shell it is OK (we have proper ulimit settings via /etc/init.d/nginx file), but despite ulimit settings in init.d Tailwatchd cannot restart Nginx, it definitely has low/default open files limit somewhere.

Perhaps it is CSH so we need to use "limit descriptors" instead of "ulimit -n"?

May someone assist?