On Wed, 2008-01-02 at 23:07 -0700, Byron Clark wrote: > On Wed, Jan 02, 2008 at 10:42:01PM -0700, Dave Smith wrote: > > I am looking for a watchdog daemon that monitors individual processes > > (given a pid file for each one), and restarts one if its pid is found to > > not be running. This is different than the typical watchdog, which would > > reboot the entire computer if a problem were found. Does this exist? I > > would also like for it to quit watchdogging a process if it continually > > crashes (i.e., if it is restarted N times in M seconds), or at least > > slow down on the restart interval. Any ideas? > > No idea if it's any good, but you may want to take a look at ps-watcher > (http://ps-watcher.sourceforge.net/).
I've used ps-watcher before and had good success with it. The config file is essentially written in Perl, which you may like or dislike. It's quite powerful although for some certain kinds of tests there are a few hoops to jump through (can't recall the details). I do remember over the years having a couple occasions where ps-watcher freaked out and kept restarting a process. Corey /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
