[ZWeb] "NZO box" on front page

[Tim Peters]

[Andrew Sawyers]
> Our postfix is fast and fine; it's that messages are not even being
> delivered to postfix from mailman due to the observations I described in
> my previous message.

The stats I attached show that m.p.o is sending out a gigabyte of email per
day, and Mailman is the only significant source of outgoing email on m.p.o
(I expect that the traffic from the comp.lang.python mailing-list gateway
alone dwarfs all of Zope's list traffic combined).  There's no backlog, and
the box isn't thrashing (although you can't tell that from the stats alone).

> I suppose we won't argue semantics - Lets just say, mailman is flakey  :)

Yes.

> In any case, when we had experienced all the problems with the commingling
> of m.z.o and m.p.o I sent a bunch of emails to postmaster at p.o
>   - to no avail.  Our best hope is someone knows mailman on high volume
> sites better then Jens or I.

I'd try again, if I were you.  Since the m.p.o postmasters are no longer
fighting any problems with m.p.o's email, (a) they may be able to explain
what they did now that it's *done*; and, (b) they have more "spare time"
now.  OTOH, if the problems were due to flakiness in a particular Zope
mailing list, perhaps they got such improvement simply by getting that list
off the m.p.o hardware.

> It could be even corruption with a single list as far as I know; I just
> don't know and have little time to spelunk m.z.o   We spent several hours
> and are certain the problem is in qfiles/out    -  I'll see if I can look
> into single mta deliveries so that groups aren't possibly being tossed
> due to a single recipient failure, as I'm guessing that might be what
> could be happening.  I'm not even certain who manages the lists; but it
> would sure help if that individual or individuals could assist.

I run, e.g., zodb-dev at zope.org.  What specifically could I do to help
determine if that list is causing you problems?