18.11. Tips and Tricks

There are a number of things you can do to make managing a sendmail site efficient. A number of management tools are provided in the sendmail package; let's look at the most important of these.

18.11.1. Managing the Mail Spool

Mail is queued in the /var/spool/mqueue directory before being transmitted. This directory is called the mail spool. The sendmail program provides a means of displaying a formatted list of all spooled mail messages and their status.

The /usr/bin/mailq command is a symbolic link to the sendmail executable and behaves indentically to:
# sendmail -bp
The output displays the message ID, its size, the time it was placed in the queue, who sent it, and a message indicating its current status. The following example shows a mail message stuck in the queue with a problem:
$ mailq Mail Queue (1 request) --Q-ID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------ RAA00275 124 Wed Dec 9 17:47 root (host map: lookup (tao.linux.org.au): deferred) terry@tao.linux.org.au
This message is still in the mail queue because the destination host IP address could not be resolved.

We can force sendmail to process the queue now by issuing the /usr/bin/runq command.

The runq command produces no output. sendmail will begin processing the mail queue in the background.

18.11.2. Forcing a Remote Host to Process its Mail Queue

If you use a temporary dial-up Internet connection with a fixed IP address and rely on an MX host to collect your mail while you are disconnected, you will find it useful to force the MX host to process its mail queue soon after you establish your connection.

A small perl program is included with the sendmail distribution that makes this simple for mail hosts that support it. The etrn script has much the same effect on a remote host as the runq command has on our own. If we invoke the command as shown in this example:
# etrn vstout.vbrew.com
we will force the host vstout.vbrew.com to process any mail queued for our local machine.

Typically you'd add this command to your PPP ip-up script so that it is executed soon after your network connection is established.

18.11.3. Analyzing Mail Statistics

sendmail collects data on mail traffic volumes and some information on hosts to which it has delivered mail. There are two commands available to display this information, mailstats, and hoststat. mailstats

The mailstats command displays statistics on the volume of mail processed by sendmail. The time at which data collection commenced is printed first, followed by a table with one row for each configured mailer and one showing a summary total of all mail. Each line presents eight items of information:

MThe mailer (transport protocol) number
msgsfrThe number of messages received from the mailer
bytes_fromThe Kbytes of mail from the mailer
msgstoThe number of messages sent to the mailer
bytes_toThe Kbytes of mail sent to the mailer
msgsregThe number of messages rejected
msgsdisThe number of messages discarded
MailerThe name of the mailer

A sample of the output of the mailstats command is shown in Example 18-5.

Example 18-5. Sample Output of the mailstats Command

# /usr/sbin/mailstats Statistics from Sun Dec 20 22:47:02 1998 M msgsfr bytes_from msgsto bytes_to msgsrej msgsdis Mailer 0 0 0K 19 515K 0 0 prog 3 33 545K 0 0K 0 0 local 5 88 972K 139 1018K 0 0 esmtp ============================================================= T 121 1517K 158 1533K 0 0

This data is collected if the StatusFile option is enabled in the sendmail.cf file and the status file exists. Typically you'd add the following to your sendmail.cf file:
# status file O StatusFile=/var/log/sendmail.st

To restart the statistics collection, you need to make the statistics file zero length:
> /var/log/sendmail.st
and restart sendmail. hoststat

The hoststat command displays information about the status of hosts that sendmail has attempted to deliver mail to. The hoststat command is equivalent to invoking sendmail as:
sendmail -bh

The output presents each host on a line of its own, and for each the time since delivery was attempted to it, and the status message received at that time.

Example 18-6 shows the sort of output you can expect from the hoststat command. Note that most of the results indicate successful delivery. The result for earthlink.net, on the other hand, indicates that delivery was unsuccessful. The status message can sometimes help determine the cause of the failure. In this case, the connection timed out, probably because the host was down or unreachable at the time delivery was attempted.

Example 18-6. Sample Output of the oststat Command

# hoststat -------------- Hostname ---------- How long ago ---------Results--------- mail.telstra.com.au 04:05:41 250 Message accepted for scooter.eye-net.com.au 81+08:32:42 250 OK id=0zTGai-0008S9-0 yarrina.connect.com.a 53+10:46:03 250 LAA09163 Message acce happy.optus.com.au 55+03:34:40 250 Mail accepted mail.zip.com.au 04:05:33 250 RAA23904 Message acce kwanon.research.canon.com.au 44+04:39:10 250 ok 911542267 qp 21186 linux.org.au 83+10:04:11 250 IAA31139 Message acce albert.aapra.org.au 00:00:12 250 VAA21968 Message acce field.medicine.adelaide.edu.au 53+10:46:03 250 ok 910742814 qp 721 copper.fuller.net 65+12:38:00 250 OAA14470 Message acce amsat.org 5+06:49:21 250 UAA07526 Message acce mail.acm.org 53+10:46:17 250 TAA25012 Message acce extmail.bigpond.com 11+04:06:20 250 ok earthlink.net 45+05:41:09 Deferred: Connection time 

The purgestat command flushes the collected host data and is equivalent to invoking sendmail as:
# sendmail -bH

The statistics will continue to grow until you purge them. You might want to periodically run the purgestat command to make it easier to search and find recent entries, especially if you have a busy site. You could put the command into a crontab file so it runs automatically, or just do it yourself occasionally.