[Zope-dev] Crash & Burn with Zope 2.6.0a1 and ZEO 2.0a1

[Stefan H. Holek]

--On Dienstag, 18. Juni 2002 11:28 -0400 Jeremy Hylton <jeremy@zope.com> 
wrote:

> I am able to run ZEO 1.0 on Zope 2.6.  Does the ZEO test suite pass
> for you?  That's spelled:

Yes, the tests pass with the two exceptions you mentioned.

> If you check the ZEO 1.0 tarball, you'll find a trigger module in it.
> Perhaps something was corrupted in your installation.

I am starting the *very same* instance that works with 2.5.1 under 2.6.0a1. 
Yes, trigger.py is there. I am stumped myself. The client starts up 
completely, but than blows up on the trigger import. This is the logfile:

2002-06-18T19:35:33 INFO(0) client Trying to connect to server: 
('127.0.0.1', 8091)
------
2002-06-18T19:35:33 INFO(0) ClientStorage Connected to storage
------
2002-06-18T19:35:33 INFO(0) ZODB Opening database for mounting: 
'142837952_1021730381.442598'
------
2002-06-18T19:35:33 INFO(0) ZODB Mounted database 
'142837952_1021730381.442598' at /temp_folder
------
2002-06-18T19:35:43 INFO(0) ZServer HTTP server started at Tue Jun 18 
19:35:43 2002
        Hostname: phoenix
        Port: 8080
Traceback (most recent call last):
  File "/Zope-2.6.0a1-src/z2.py", line 831, in ?
    asyncore.loop()
  File "/Zope-2.6.0a1-src/lib/python/ThreadedAsync/LoopCallback.py", line 
100, in loop
    _start_loop(map)
  File "/Zope-2.6.0a1-src/lib/python/ThreadedAsync/LoopCallback.py", line 
70, in _start_loop
    apply(cb, (map,) + args, kw or {})
  File "/home/zeo6/lib/python/ZEO/ClientStorage.py", line 215, in 
becomeAsync
  File "/home/zeo6/lib/python/ZEO/ClientStorage.py", line 555, in getWakeup
ImportError: No module named trigger

I am going to investigate whether this is an artefact of my instance but I 
am not sure where to start...

> I don't quite follow.  Can you explain exactly what went wrong when
> you tried to run the server and client from the same directory?  I
> assume this is a problem with ZEO 2.0.  What do you mean by "the new
> startup stuff"?

      - Collector #1: UNIX security fixes: make starting Zope as 'root'
        secure, stop using 'nobody', warn of insecure umasks

Things like setting the sticky bit on var/, chdir to var/, umask 077 in the 
start file, creating the pid file *before* dropping root. Maybe more.

When running ZEO 1.0 Data.fs.lock (and Data.fs.tmp) cannot be owned by 
root. ZEO_SERVER.pid is owned by the unprivileged user as well. Ok, "many" 
is overstated, I admit. ;-)

Thanks,
Stefan

--
BLOWFISH, n. - Preference for beef.