This topic seems to come up from time to time. I was very surprised when
I moved to nginx to find out that it is very limited in its
self-monitoring / restoration approach when used as a load balancer.
Nginx is fantastic and very robust seemingly in every way except this
area. Unfortunately it is an important area.
A while back Francois (I think?) was talking about making a module to
make for integrated centralized reporting of status for nginx that could
then be hooked into. It then feel silent. I hope to see something like
this emerge from these talks but always seems to disapate after a while
and we are left with bandaid appraoches instead of a true solution.