1MIMEDEFANG-MULTIPLEXOR(8) System Manager's Manual MIMEDEFANG-MULTIPLEXOR(8)
2
6 mimedefang-multiplexor - Process pool controller for mail filters.
7
9 mimedefang-multiplexor [options]
10
13 mimedefang-multiplexor manages a pool of Perl processes for scanning e-
14 mail. It is designed to work in conjunction with mimedefang(8) and
15 mimedefang.pl(8).
16 mimedefang-multiplexor opens a UNIX-domain socket and listens for
18 requests for work from mimedefang. As requests come in, mimedefang-
19 multiplexor creates Perl processes as needed to scan mail. The Perl
20 processes are not killed when scanning is completed, but continue to
21 run in a loop. Perl processes are re-used for subsequent e-mail mes‐
22 sages. This eliminates the large overhead of starting a new Perl
23 process for each incoming message.
24 To avoid memory leaks, the Perl processes are killed after they have
26 handled some number of scans.
27
30 -U user
31 Runs the multiplexor as user rather than root. This option is
32 mandatory, and must match the -U option supplied to mimedefang.
33 -m minWorkers
36 The minimum number of Perl processes to keep running at all
37 times. The default is zero.
38 -x maxWorkers
41 The maximum number of Perl processes to run simultaneously. If
42 a request comes in and all processes are busy, a temporary fail‐
43 ure is signalled to the SMTP peer. The default is 2.
44 -r maxRequests
47 The maximum number of requests a given process handles before it
48 is killed and a replacement started. The default is 500.
49 -i idleTime
52 The idle time in seconds after which to kill of excess Perl pro‐
53 cesses. That is, if the process is idle for longer than this
54 time, and there are more than minWorkers running, the process is
55 killed. Note that this is implemented as a timer which ticks
56 every idleTime seconds; therefore, processes may be idle for up
57 to twice this time before they are killed. The default for
58 idleTime is 300 seconds.
59 -V maxLifetime
62 The maximum lifetime in seconds of a worker before it is killed
63 and a replacement started. The default is -1, which signifies
64 no maximum lifetime. Note that the lifetime check is done only
65 when a worker becomes idle after a request, or every time the
66 idle-timeout check is made. On a very quiet system, workers may
67 live for longer than maxLifetime by as much as idleTime. Note
68 also that the lifetime is measured not from when the worker
69 started running, but from when it was assigned its very first
70 request. A completely-idle worker that has never processed any
71 requests will not be terminated by the maxLifetime setting.
72 -b busyTime
75 The longest a Perl process is allowed to spend scanning an e-
76 mail before it is declared hung up and killed. The default is
77 120 seconds.
78 -Z This option specifies that the multiplexor should accept and
81 process "status updates" from busy workers. Note that this con‐
82 sumes one extra file descriptor per worker, plus a small amount
83 of CPU time per status update.
84 -c cmdTime
87 The timeout for communication between mimedefang-multiplexor and
88 mimedefang, or between mimedefang-multiplexor and a Perl scan‐
89 ning process. The default is 10 seconds. This timeout should
90 be kept quite short.
91 -w waitTime
94 When mimedefang-multiplexor starts the initial workers, or needs
95 to bring the number of running workers up to the number defined
96 by the -m option, it does not start all the workers at once,
97 because this could overload your server. Instead, it starts one
98 worker every waitTime seconds. The default value for waitTime
99 is 3.
100 -W waitTime
103 If you use this option, mimedefang-multiplexor will never acti‐
104 vate a worker until waitTime seconds have elapsed since the last
105 worker activation. This could result in mail being tempfailed
106 if worker activations do not keep pace with incoming mail. How‐
107 ever, it may be preferable to tempfail mail rather than allow
108 the load on your server to spike up too quickly. The default
109 value for this option is 0, meaning that mimedefang-multiplexor
110 will start workers as quickly as necessary to keep up with
111 incoming mail.
112 -z spooldir
115 Set the spool directory to spooldir. If this option is omitted,
116 the spool directory defaults to /var/spool/MIMEDefang.
117 -s pathName
120 The UNIX-domain socket on which mimedefang-multiplexor listens
121 for requests. This should be specified as an absolute pathname.
122 If this option is not supplied, it defaults to mimedefang-multi‐
123 plexor.sock under the spool directory.
124 -a socket
127 A socket for listening for requests. This is similar to the -s
128 socket, except that a restricted set of requests are processed.
129 On this socket, the multiplexor will only process requests ask‐
130 ing for status; it will not accept any commands to do scanning
131 or that would consume a worker. See the SOCKET SPECIFICATION
132 section for the format of socket.
133 -p fileName
136 Causes mimedefang-multiplexor to write its process-ID (after
137 becoming a daemon) to the specified file. The file will be
138 owned by root.
139 -o fileName
142 Causes mimedefang-multiplexor to use fileName as a lock file to
143 avoid multiple instances from running. If you supply -p but not
144 -o, then mimedefang-multiplexor constructs a lock file by
145 appending ".lock" to the pid file. However, this is less secure
146 than having a root-owned pid file in a root-owned directory and
147 a lock file writable by the user named by the -U option. (The
148 lock file must be writable by the -U user.)
149 -f filter_path
152 Normally, mimedefang-multiplexor executes a Perl filter script
153 called mimedefang.pl to scan the e-mail. However, you can have
154 it execute any program you like by specifying the full path to
155 the program with the -f option. This program must obey the pro‐
156 tocol documented in mimedefang-protocol(7); see that manual page
157 for details.
158 Note that the -f option does not specify the "filter" to use
160 with mimedefang.pl; instead, it specifies the program for
161 mimedefang-multiplexor to execute. You almost certainly should
162 not use this option unless you wish to replace mimedefang.pl
163 with your own program.
164 -F rules_path
167 Specifies the path to the filter rules. By default,
168 /etc/mail/mimedefang-filter is used. If you use the -F option,
169 its value is passed to the underlying Perl filter program using
170 -f.
171 -l Log certain events, including the output of the Perl workers'
174 standard-error, using syslog. Normally, the multiplexor does
175 not log much information.
176 -d Write debugging information about event-handling code in
179 /var/log/mimedefang-event-debug.log. This is only of use to
180 people debugging mimedefang-multiplexor.
181 -R kbytes
184 Limits the resident-set size of the worker filter processes to
185 kbytes kilobytes. This limit is not supported on all operating
186 systems; it is known to work on Linux.
187 -M kbytes
190 Limits the total memory space of worker filter processes to
191 kbytes kilobytes. This limit is supported on all operating sys‐
192 tems which support the setrlimit(2) system call. This should
193 include most modern UNIX systems.
194 We recommend that you monitor your worker filter processes and
196 get a feel for how much memory they use. You should then limit
197 the memory to two or three times the worst-case that you have
198 observed. This can help mitigate denial-of-service attacks
199 which use complicated MIME messages to force mimedefang.pl to
200 consume lots of memory.
201 -h Print usage information and exit.
204 -t filename
207 Log statistical information to filename. See the section STA‐
208 TISTICS for more information.
209 -T Log statistical information using syslog(2). You may use any -t
212 and -T together, in which case statistical information is logged
213 in a file and using syslog.
214 -u Flush the statistics file after every write. Normally, mimede‐
217 fang-multiplexor does not flush the file; this is the best
218 choice for minimizing disk I/O on a busy mail server. However,
219 if you wish to watch statistics entries in real-time, you should
220 enable flushing.
221 -D Do not fork into the background and become a daemon. Instead,
224 stay in the foreground. Useful mainly for debugging or if you
225 have a supervisory process managing mimedefang-multiplexor.
226 -q queue_size
229 Normally, if all workers are busy and mimedefang-multiplexor
230 receives another request, it fails it with the error "No free
231 workers." However, if you use the -q option, then up to
232 queue_size requests will be queued. As soon as a worker becomes
233 free, the queued requests will be handed off in FIFO order. If
234 the queue is full and another request comes in, then the request
235 is failed with "No free workers".
236 -Q queue_timeout
239 Queued requests should not stay on the queue indefinitely. If a
240 queued request cannot be processed within queue_timeout (default
241 30) seconds of being placed on the queue, it is failed with a
242 "Queued request timed out" message. See the section "QUEUEING
243 REQUESTS" for more discussion.
244 -O sock
247 Listen on a notification socket for connections from listeners.
248 mimedefang-multiplexor can inform external programs of state
249 changes by sending messages over a notification socket. The
250 external programs connect to this socket and then listen for
251 notifications. See the section SOCKET SPECIFICATION for the
252 format of sock.
253 See the mimedefang-notify(7) man page for details of the notifi‐
255 cation protocol.
256 -N map_sock
259 Listen on a map socket for Sendmail SOCKETMAP connections. As
260 of Sendmail 8.13, you can define a Sendmail map type that talks
261 to a daemon over a socket. mimedefang-multiplexor implements
262 that protocol; consult the mimedefang-filter(5) man page for
263 detils (see the SOCKET MAPS section).
264 See the section SOCKET SPECIFICATION for the format of map_sock.
266 -I backlog
271 When mimedefang-multiplexor creates a listening socket, it cal‐
272 culates the "backlog" argument to listen(2) based on the maximum
273 number of workers. However, you can explicitly set this backlog
274 with the -I option. Setting the backlog to a high value (around
275 30-50) may help on a very busy server. If you see mail log mes‐
276 sages saying "MXCommand: socket: Connection refused" during busy
277 periods, then that's an indication you need a higher listen
278 backlog.
279 -L interval
282 Log the worker status every interval seconds. This logs a line
283 using syslog; the line looks like this:
284 Worker status: Stopped=s Idle=i Busy=b Killed=k Queued=q Msgs=m Activations=a
287 Here, "Stopped" is the number of non-running workers, "Idle" is
289 the number of idle workers, "Busy" is the number of busy work‐
290 ers, "Killed" is the number of killed workers yet to be reaped,
291 "Queued" is the number of queued requests, "Msgs" is the total
292 number of messages processed since the multiplexor began run‐
293 ning, and "Activations" is the number of times a Perl process
294 has been started since the multiplexor began running.
295 If you supply an interval of 0 (which is the default), no peri‐
297 odic status logging is performed. If you supply an interval of
298 less than 5 seconds, it is silently reset to 5 seconds.
299 -S facility
302 Specifies the syslog facility for log messages. The default is
303 mail. See openlog(3) for a list of valid facilities. You can
304 use either the short name ("mail") or long name ("LOG_MAIL") for
305 the facility name.
306 -E Specifies that the multiplexor should create an embedded Perl
309 interpreter. This can improve performance dramatically. But
310 see the section "EMBEDDING PERL" for more information.
311 -X n Specifies that the multiplexor should initiate a "tick" request
314 every n seconds. This causes your filter_tick function (if
315 defined) to be called. Note that you have no control over which
316 worker executes filter_tick. If all workers are busy when a
317 tick occurs, that tick request is skipped and a warning message
318 is logged.
319 -P n Specifies that the multiplexor should run n tick requests in
322 parallel. Each tick is run as often as specified with the -X
323 argument. (If you omit the -P option, then the multiplexor
324 behaves as if -P 1 had been specified.)
325 If you run parallel ticks, each tick is assigned an integer
327 identifying its "type". The type ranges from 0 to n-1. While
328 there may be as many as n tick requests running at a time, only
329 one tick of each type will be active at any time.
330 -Y label
333 Sets the tag used in the multiplexor's syslog messages to label
334 instead of mimedefang-multiplexor.
335 -G Normally, mimedefang-multiplexor uses a umask of 027 when creat‐
338 ing listening sockets. If you would like the sockets to be
339 readable and writeable by the group as well as the owner, supply
340 the -G option. This causes the umask to be 007 whenever UNIX-
341 domain sockets are created.
342 -y n Limits the maximum number of concurrent recipok checks to n on a
345 per-domain basis. The value of n can range from 0 (in which
346 case no limit is applied) to maxWorkers, where maxWorkers is the
347 argument to the -x option. If n is outside that range, it is
348 ignored (and no limit is applied.)
349 The recipok command ultimately invokes the filter_recipient
352 function in your filter. If you are doing recipient verifica‐
353 tion against servers that may be slow or unreliable, you can use
354 the -y option to limit the number of concurrent recipient veri‐
355 fications per domain. That way, if one domain's server becomes
356 very slow, it won't consume all available workers for recipient
357 verification. Instead, its RCPT commands will be tempfailed and
358 there will be workers available to handle RCPT commands for
359 other domains.
360
363 The -a, -N and -O options take a socket as an argument. This socket
364 can be specified as:
365 /path/to/socket
368 A UNIX-domain socket
369 inet:portnum
372 A TCP socket bound to port portnum, but which accepts connec‐
373 tions only from the IPv4 loopback address (127.0.0.1).
374 inet_any:portnum
377 A TCP socket bound to port portnum which will accept connections
378 from any address. Use inet_any with caution!
379 inet6:portnum
382 A TCP socket bound to port portnum listening on the IPv6 loop‐
383 back address.
384 inet6_any:portnum
387 A TCP socket bound to port portnum listening on the IPv6 wild‐
388 card address.
389
392 Normally, if all workers are busy, any additional requests are failed
393 immediately. However, the -q and -Q options allow you to queue
394 requests for a short amount of time. This facility is intended to
395 gracefully handle a temporary overload; most of the time, your queue
396 should be empty.
397 Because mimedefang checks the number of free workers when a connection
399 is opened and fails the connection if there are no free workers, the
400 intent of the queue is to allow SMTP transactions that are already
401 underway to continue if there is a slight overload. Any new connec‐
402 tions will be failed if all workers are busy, but existing connections
403 are allowed to continue. Queuing requests may improve throughput on
404 extremely busy servers.
405 Note that if you supply the -q option to mimedefang, then even new con‐
407 nections are allowed to queue. This may improve throughput by keeping
408 the worker utilization higher.
409 The -R option to mimedefang can be used to reserve a specified number
411 of workers for connections from the loopback address. Using the -R
412 option has the side-effect of permitting new connections from the loop‐
413 back address to queue.
414
417 Normally, when mimedefang-multiplexor activates a worker, it forks and
418 execs mimedefang.pl. However, if the multiplexor was compiled with
419 embedded Perl support, and you supply the -E command-line option, the
420 multiplexor works like this:
421 1 It creates an embedded Perl interpreter, and sources mimede‐
424 fang.pl with a special command-line argument telling it to read
425 the filter, but not to enter the main loop.
426 2 Each time a worker is activated, the multiplexor calls fork()
429 and runs the mimedefang.pl main loop. This invokes filter_ini‐
430 tialize and then runs the main loop.
431 On some platforms (for example, Red Hat Linux 7.3 with Perl 5.6.1), it
434 is not safe to destroy and recreate a Perl interpreter without causing
435 a memory leak. On those platforms, if you attempt to reread the filter
436 file (by sending the multiplexor a HUP signal or reread command), the
437 filter will not be re-read, and a message will be logged to syslog. On
438 those platforms, you must kill and restart MIMEDefang if you change the
439 filter file.
440 On most platforms, however, a filter reread is accomplished by destroy‐
443 ing and re-creating the embedded interpreter, re-sourcing mimedefang.pl
444 and killing workers as soon as they are idle.
445
448 With the -t option, mimedefang-multiplexor logs certain events to a
449 file. This file can be post-processed to gather statistics about the
450 multiplexor. You can use it to tune the number of workers you run,
451 adjust timeouts, and so on.
452 Each line of the file looks like this:
455 YYYY/MM/DD:HH:MM:SS timestamp event key=val key=val...
457 Here, YYYY/MM/DD:HH:MM:SS is the local time of day. Timestamp is the
459 number of seconds since January 1, 1970. Event is the name of an
460 event. The valid events are:
461 StartWorker
464 A worker process has been started.
465 KillWorker
468 A worker process has been killed.
469 ReapWorker
472 A dead worker process has been reaped. It is possible to have a
473 ReapWorker event without a previous KillWorker event if the
474 worker process terminated abnormally.
475 StartFilter
478 A worker process has begun filtering an e-mail message.
479 EndFilter
482 A worker process has finished filtering an e-mail message.
483 The possible keys in the key=value pairs are:
486 worker=n
489 The worker involved in the event. Every worker is identified by
490 a small integer.
491 nworkers=n
494 The total number of running workers immediately after the event
495 happened.
496 nbusy=n
499 The number of busy workers (workers which are processing an e-
500 mail message) immediately after the event happened.
501 reason="string"
504 The reason for a StartWorker or KillWorker event. (Present only
505 for these events.)
506 numRequests=n
509 The number of e-mails processed by the worker. Present only for
510 an EndFilter event.
511 If you send the mimedefang-multiplexor process a SIGHUP signal (kill -1
514 pid), it closes and reopens the statistics file. This is useful during
515 log file rotation.
516 If you send the mimedefang-multiplexor process a SIGINT signal (kill
519 -INT pid), it terminates all active-but-idle workers. Also, any
520 active-and-busy workers terminate as soon as they finish filtering the
521 current message. This is useful to force a reread of the filter rules
522 file without stopping and restarting Sendmail.
523 If you send the mimedefang-multiplexor process a SIGTERM signal (kill
526 pid), it terminates all workers and exits immediately.
527
530 mimedefang-mulitplexor was written by Dianne Skoll <dfs@roaringpen‐
531 guin.com>. The mimedefang home page is http://www.mimedefang.org/.
532
535 mimedefang.pl(8), mimedefang-filter(5), mimedefang(8), mimedefang-pro‐
536 tocol(7)
5374th Berkeley Distribution 8 February 2005 MIMEDEFANG-MULTIPLEXOR(8)