The listener processes become zombied and stop responding to requests. The
OAS failure recovery mechansim will kill and restart the listener process
automatically (if OAS is using the non-collapsed model) 80% of the time.
When that fails owsctl reload -l listenername usually will restart the
listener.
Joe Kazimierczyk <joseph.kazimierczyk_at_bms.com> wrote in message
news:38296670.BD271CAC_at_bms.com...
> When you say the listeners are failing, what exactly do you mean?
>
>
> Ann Downey wrote:
> >
> > We're running OAS 4.0.7.1 on NT 4.0 SP3 on a very busy site. We have to
> > balance load across several servers, each with 2 listeners. We noticed
that
> > the listeners were failing at a rate proportional to the hits/minute
being
> > received. When we adjusted the load balancing to send traffic to the
> > different listeners with the ratios of 1:2:3:5:7:11 the failures
occurred on
> > each listener at rates proportional to the load ratios.
> >
> > Clearly the listeners are failing when some threshold or counter or
register
> > is exceeded. What could this be?
> >
> > Thanks in advance.
> >
> > Ken
Received on Wed Nov 10 1999 - 21:50:18 CST
