Earlier this week, I started seeing this error on a FreeBSD 9.2 mail server running Postfix 2.10.1:
CDDB71C56D53 37470 Mon Nov 11 16:22:45 MAILER-DAEMON
(temporary failure. Command output: maildrop: signal 0x19)
mailarchives@tallboy.unixathome.org
I use maildrop, a procmail alternative. I first started using procmail in December 2000. It’s been a good 13 years, but it’s time. At the prompting of Warren Block, who has helped me (and many others) on numerous occasions in the FreeBSD Forums, I started looking at it. The development site for FreshPorts is now running on maildrop.
But back to the issue at hand.
It took me a while to locate the cause of the problem:
0x19 is likely SIGXFSZ (file size exceeded) and likely your mailbox
files are 50,000 kBytes large which is the default file size limit that
Postfix imposes, figure your current limits with: …
Ahh!
Checking my server I found:
# postconf | grep -E 'm.*_size_limit' mailbox_size_limit = 51200000 message_size_limit = 32768000
That’s roughly 48M
Checking the mailboxes:
# ls -lh mail total 89960 -rw------- 1 dan dan 48M Nov 11 17:12 mbox -rw------- 1 dan dan 1.2M Nov 11 00:00 mbox.0.gz -rw------- 1 dan dan 1.1M Nov 10 00:00 mbox.1.gz -rw------- 1 dan dan 2.0M Nov 1 00:00 mbox.10.gz
You might wonder why I have mailboxes which look like that?
According to the above, I have clearly hit the limit. Let’s alter /usr/local/etc/postfix/main.cf and set:
mailbox_size_limit = 512000000
Now to restart postfix:
service postfix restart
And get a push on those waiting emails:
postfix flush
After the mail clear through, my files looked like this:
# ls -lh mail total 130408 -rw------- 1 mailarchives mailarchives 88M Nov 11 17:14 mbox -rw------- 1 mailarchives mailarchives 1.2M Nov 11 00:00 mbox.0.gz -rw------- 1 mailarchives mailarchives 1.1M Nov 10 00:00 mbox.1.gz
All good now!
