[Zope] Zope hangs during load (Panic!)

[Petter Holmström]

Hello,

We have a strange and very irritating problem with our Zope/Plone-site, 
and I'd like to know where the problem is.

It all started a couple of days ago when we were running Zope 2.6.4. An 
upgrade to version 2.7.0 did not solve it, which could mean that the 
problem is to be found in some of the Plone-products, couldn't it?

First, I start Zope with the ./runzope-command. Debug-mode is off, max 
threads is 6. Proxy-routing and SSL via Apache2. FreeBSD 5.2.1 with 
Python 2.3.3 (WANT_HUGE_STACK_SIZE=1).
Products:
- Plone 2.0RC5
- CMF 1.4.2
- Epoz 0.7.4
- CMFMessage 1.1
- ZWiki 0.27.1
- Archetypes 1.2.3
- PortalTransforms 1.0.2
- CMFBoard 1.4.2

top-output (the command-column is always python):

   PID USERNAME PRI NICE   SIZE    RES STATE  C   TIME   WCPU    CPU 
92459 zope      96    0 47348K 45152K select 0   0:12  0.00%  0.00%

Then, I run a benchmark-test:

# http_load -rate 5 -seconds 10 urls
2 fetches, 48 max parallel, 40332 bytes, in 10.7343 seconds
20166 mean bytes/connection
0.186319 fetches/sec, 3757.32 bytes/sec
msecs/connect: 3.0435 mean, 5.481 max, 0.606 min
msecs/first-response: 815.624 mean, 1172.58 max, 458.665 min
HTTP response codes:
   code 200 -- 2

Now top-output looks like this:

   PID USERNAME PRI NICE   SIZE    RES STATE  C   TIME   WCPU    CPU 
92459 zope      96    0 55668K 53768K select 0   0:15  1.22%  1.22% 
92499 zope      96    0 55412K 53552K select 0   0:00  0.00%  0.00% 
92500 zope      96    0 55668K 53604K select 0   0:00  0.00%  0.00% 
92495 zope      96    0 52504K 50596K select 1   0:00  0.00%  0.00% 
92498 zope      96    0 55180K 53128K select 0   0:00  0.00%  0.00% 
92491 zope      96    0 49220K 47012K select 0   0:00  0.00%  0.00% 
92493 zope      96    0 49220K 47144K select 0   0:00  0.00%  0.00%

When I now try to access the zope-server via a webbrowser nothing 
happens, I just get a "Waiting for [hostname]...." until a time-out occures.

Same benchmark-test:

# http_load -rate 5 -seconds 10 urls
0 fetches, 50 max parallel, 0 bytes, in 10.4793 seconds
0 fetches/sec, 0 bytes/sec
HTTP response codes:

I stop the runzope-process with Ctrl-C. Top-output after this:

   PID USERNAME PRI NICE   SIZE    RES STATE  C   TIME   WCPU    CPU 
92499 zope      96    0 55412K 53552K select 0   0:00  0.00%  0.00% 
92495 zope      96    0 52504K 50596K select 1   0:00  0.00%  0.00% 
92500 zope      96    0 55668K 53604K select 1   0:00  0.00%  0.00% 
92498 zope      96    0 55180K 53128K select 0   0:00  0.00%  0.00% 
92491 zope      96    0 49220K 47012K select 0   0:00  0.00%  0.00% 
92493 zope      96    0 49220K 47144K select 1   0:00  0.00%  0.00%

These processes must be killed with signal 9.

The error-logs do not contain anything, nor does the runzope-program 
output anything strange. The same thing happens if I run zope in debug-mode.

Please help! We need this server in production use, and if we can't get 
it to work we have to replace Zope with something else, and I don't know 
with what...

Thanks in advance,

-Petter-