[Deploying Sakai] mod_proxy_balancer and nofailover=On

[Martin B. Smith]

Howdy Production Sakai community,

I'm hoping to get some comments on using Apache front ends routed to
Tomcat instances using mod_proxy_balancer. I notice that every single
example (including the Sakai Admin guide's "Advanced Tomcat" section)
recommends a line like:

ProxyPass / balancer://mycluster/ stickysession=JSESSIONID|jsessionid
nofailover=On

Apache's documentation for the "nofailover" setting:
"set to On the session will break if the worker is in error state or
disabled. Set this value to On if backend servers do not support session
replication."

I'm noticing that if I actually leave nofailover=Off, when I shutdown a
tomcat instance, switching to the new Tomcat node doesn't appear to
create a major problem. I'm not doing anything interesting with my
session, though, so perhaps I just don't know the particular failure
scenario we're avoiding with "nofailover=On".

With some creative Googling, I *do* find examples of folks using
"nofailover=Off" in production with Sakai, mostly from list archives.

If you *are* running a system with mod_proxy_balancer and
"nofailover=On", how do you move users off of a particular node to do
maintenance? Do you use the balancer-manager feature? If you're running
a system with "nofailover=Off" or with the setting omitted entirely, do
you encounter any issues with failing users over from one tomcat to another?

Thanks in advance for the discussion! :)
-- 
Martin B. Smith
smithmb at ufl.edu - (352) 273-1374
CNS/Open Systems Group
University of Florida