[Zope] PCGI/Zope instabilities.

[John Eikenberry]

Hello,

We are using Zope for some of the functionality on a high traffic (150K
hits a day) site we maintain. It's turning out to be very unstable (on our
platform at least). I'd like some advice on how to get it more stable
(besides switching hardware/platforms, which is on the agenda down the
road).

What happens is that somehow a bunch of additional pcgi_publisher.py
processes get started until eventually the zope app is unreachable
("Temporarily Unavailable" errors). 

I've seen up to around 10 of these additional python processes running at
the same time. To get things going again, I have to login and kill them
all and restart Zope (by reloading it - I have to do this about once
every day or two). 

In the pcgi.log file, I find a ton of these messages:

Thu Apr 29 11:02:57 1999
  pcgi-wrapper: pcgiVerifyProc failed  (103) unable to verify if process
is running

I'm guessing that mabey the zope process isn't returing fast enough for
the pcgi process, which then thinks it needs to start up zope. Or
something like this.

My initial idea is to modify pcgi's timeout, to see if making it a little
longer would help. 

Please feel free with any advice/tips/etc.

Thanks,

PS. some specs:

Zope-1.10.2
Apache-1.3.3
BSDI 3.1 (strange setup with multiple virtual machines on one physical
		machine - they claim it's no different than a full system)
python-1.5.1 (all modules are statically compiled in - including zope
		modules)

---

John Eikenberry
[jae@kavi.com - http://taos.kavi.com/~jae/] 
______________________________________________________________
"A society that will trade a little liberty for a little order
 will deserve neither and lose both."
                                         --B. Franklin