> Kevin Sawyer <[EMAIL PROTECTED]> wrote:
>
> You're seeing messages in the queue that qmail isn't attempting to
> deliver? I can think of two possible causes for that in a properly
> functioning installation:
>
> 1) concurrencyremote/local are being reached, or
> 2) the target host for a remote delivery is marked as timing-out
> in qmail-remote's list of nonresponding hosts. (see "man
> qmail-tcpto").
I'm not actually "seeing" messages really. Here's some more information.
If wait a day or so then run "du -s /var/qmail/queue" I'll see say 120MB in
that hierarchy. If I then run "/etc/rc.d/init.d/qmail restart" I can
immediately use "df" to watch my /var partition usage shrink. After 10
minutes or so it will level out and running "du -s /var/qmail/queue" will
show that the queue contents have dropped by about 50%. Next, I get dial-up
subscribers calling saying, "I just got about 40 messages that were three
days old. What's up with that?" So, it isn't just remote messages due to
non-responding hosts. It's local messages.
Before I do all of this, I notice that qmail-queue and qmail-send are
definitely running and a lot of local and remote deliveries are succeeding
according to the logs. My concurrencylocal/remote are both set to the
compiled maximum of 120 and this box is definitely mean enough to support
that. I have about 5,000 mailboxes. My local delivery agent is procmail
with Maildir patches.
It just doesn't make sense to me that only SOME messages don't get delivered
locally while others continue to get delivered just fine.
More ideas? Thanks...
--Kevin
