|
Alex Kramarov wrote:
> it's
> impossible ot monitor the service running on the vip from the same machine
> (or so it seems from the docs),
from the director, a mon monitor can't check a service on the VIP
of a realserver. What is done currently for the configure script
is to have the service listen on the RIP as well and you monitor
the service on the RIP. You hope that the functionality of the
service on the VIP and the RIP track each other. This isn't a
good solution but works for the moment.
Better would be to remotely run the monitor on the realserver and have it
check the VIP. I haven't written this into the configure script yet.
> has anyone seen a case where lvs
> fails (and I am not talking about the real servers, but about the load
> balancer module) but the machine lives on, and if such a case exist, does
> anyone has a solution for monitoring and taking action when such failure
> occures ?
no one has described, on this list, an actual director failure in an HA'ed
production system. Most people seem to check out the HA functionality by
hitting switches or pulling cables and then put it on line. What happens
after that I don't know. I've quizzed a small and statistically invalid
sample of people personally about this and it would seem that the HA
features are used for keeping managers assured that the machines are
redundant and for keeping the LVS up during planned maintenance.
Anyone got any director failure stories in HA'ed LVSs?
Joe
--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@xxxxxxx ph# 919-541-0007, RTP, NC, USA
|
