Nginx + fair load balancer patch looping

[Grzegorz Nosek]

On Fri, Mar 14, 2008 at 12:29:41AM +0100, Alexander Staubo wrote:
> Unfortunately I have no idea how to reproduce it. If the problem rears
> its head again, is there anything I should do in terms of probing the
> running Nginx process? Would a few minutes of running strace help, for
> example?

If it doesn't kill your I/O, it might be useful (as an insight into WTF
the fair balancer is choosing). I think that the most useful would be
capturing debug_http logs (only those with [upstream_fair] should be
enough), but that generates a truly massive amount of data.

A 'strace -e connect' should contain some useful data too, without
(hopefully) saturating the disk write bandwidth.

You could try the latest snapshot (or just apply
http://git.localdomain.pl/?p=nginx.git;a=commitdiff;h=e75e0a9c8cbad9b8f43848becbcc4d425cba3eae
by hand, it's a one-liner). It should fix the problem, at least regarding
dead backends. I have no idea really, how a clock skew comes into play
here.

Best regards,
 Grzegorz Nosek