[ZODB-Dev] zeo 2: asyncore?

[Mark McEahern]

[Guido van Rossum]
> This is a known bug in the ZEO 2.0b1 release (as are the other
> occasionally hanging tests you report below).  This hang is
> particularly persistent on Red Hat 7.2.  It's fixed in ZEO CVS AFAICT.
> We'll do a b2 release in the next couple of weeks.

I'll try ZEO HEAD.  (Soon as I figure out how to access the CVS, which I
don't imagine will be difficult.)

> > Hmm, now I'm beginning to wonder whether my fake TLD for internal
> > addresses is the problem.  My dhcp server is also running dnscache
> > (from djbdns) to serve these names.  What's weird about that is Zope
> > without ZEO works just fine (i.e., if I don't place the
> > custom_zodb.py in instance home).  So maybe this is a red herring.
>
> But maybe it isn't.

I removed the entry for this computer from tinydns and restarted both
tinydns and dnscache.  I then modified $Zope/start to use:

  -w 127.0.0.1:8080

in the arguments for z2.py.  It read localhost.localdomain from /etc/hosts.
No goofy .internal TLD.  And I got the same problem.  :(

Thanks,

// mark

-