keeping dccm, dccifd, and dccd running

Vernon Schryver vjs@calcite.rhyolite.com
Wed Mar 30 23:31:26 UTC 2005


I've been told that Postfix does not have an equivalent to the sendmail
mechanism of optionally ignoring a filter process (milter) that has died.
That and the current dccifd crashing every several hours or days was
given as a compelling reason to not use dccifd as before-queue filter
with Postfix.

The obvious fix for such hassles is a shell script or one of the
process monitoring applications that runs daemons with -b and restarts
them if they every quit.  Such fixes have usually have problems when
you really do want to stop the daemon.

So would it be worthwhile to add "-P rlimit" to all three daemons
to cause them to restart themselves if they ever stop unexpectedly?
(rlimit=seconds to delay consecutive restarts for rate limiting).

Or should something like -P5 be the default, 
with -P0 to turn off the restart mechanism?

I don't like the smell of such mechanisms, because I've seen people
use them as crutches to escape fixing whatever causes the real problems.
On the other hand, they're better than shell scripts that check and
restart processes.
On the third hand, they're not substitutes for shell scripts that 
check to see that the process is not only still running but working.


Vernon Schryver    vjs@rhyolite.com



More information about the DCC mailing list

Contact vjs@rhyolite.com by mail or use the form.