[Mailman-Users] Counting messages that went to postfix queue

[Mark Sapiro]

D G Teed wrote:
>
>Actually it was there.  Sorry about that.  I was looking at the wrong date
>since I was analyzing this the next day...
>
>Apr 30 15:43:24 2007 (2110) <
>EF80AA494809654B9EE67C4DD650CC7807E443AE at exchange.mydomain> smtp for 1890
>recips, completed in 2.463 seconds
>
>(domain obscured to protect the innocent)
>
>So it thought it was sending out 1890, but in postfix there are certainly
>20 some missing.


Assuming there is nothing in smtp-failure, Postfix appeared to accept
all 1890 recipients.


>These are the overrides I have:
>DEFAULT_ARCHIVE = Off
>DEFAULT_ARCHIVE_PRIVATE = 1
>ARCHIVE_TO_MBOX = 0
>VERP_CONFIRMATIONS = Yes
>DEFAULT_LIST_ADVERTISED = No
>DEFAULT_MAX_MESSAGE_SIZE = 60           # KB
>DEFAULT_SUBJECT_PREFIX  = "[%(real_name)s] "
>DEFAULT_MSG_HEADER = ""
>DEFAULT_MSG_FOOTER = ""
>DEFAULT_SEND_REMINDERS = No
>DEFAULT_SEND_WELCOME_MSG = No
>DEFAULT_FIRST_STRIP_REPLY_TO = Yes
>DEFAULT_SUBSCRIBE_POLICY = 3
>DEFAULT_UNSUBSCRIBE_POLICY = 1
>PENDING_REQUEST_LIFE = days(7)
>
>Nothing from the set you mentioned.  But anyway, as the lost log entry
>was found, I don't have any issue in this regard I don't think.


Right. My question was only related to the 'missing' log entry.


>There is only one bad entry - with the quote inside of it.
>I'll run the list through a checker before adding to lists
>in the future.  The -i flag doesn't detect that problem,
>which isn't too surprising since nothing stopped the address
>from being added.


Mailman does a few tests on addresses, but it could do better.


>We are running a second list test soon and I'll see if
>it works more robustly with data checks prior to subscribing the list
>members.


Given the mm_cfg.py overrides above (actually, some things that aren't
overridden and still have their default values), I can tell you that
Mailman will deliver to the 1890 recipients in four smtp transactions
with 500, 500, 500 and 390 recipients. The recipients will be grouped
by top level domain, but other than that are in an essentially random
order so without actually going through the 'chunking' process with
the actual addresses, we don't know in which group or where within
that group the 'bad' address appears.

I suppose it is possible that Postfix gets one of these chunks,
encounders the 'bad' address and drops the rest of the chunk, although
I would expect this to be reported back to Mailman and logged in
smtp-failure and also logged in Postfix's logs.

I suppose it is also possible and perhaps more likely that Postfix gets
the chunk with the bad address and does its own sort by domain stuff
for its delivery process and then drops a group with the 'bad'
address, although I would expect at least the 'bad' address to be
logged in Postfix's logs.

-- 
Mark Sapiro <msapiro at value.net>       The highway is for gamblers,
San Francisco Bay Area, California    better use your sense - B. Dylan