James Carlson wrote: > David Finberg writes: > >> On Tue, 6 May 2008, Robert Sturrock wrote: >> >> >>> The Hylafax startup forks _two_ daemons: faxq and hfaxd. >>> >>> "faxq" dies from time to time - we would like to use SMF to >>> automatically restart this. However, our (naive) configuration does >>> not cater for this situation as SMF won't stop/start the service as a >>> whole unless _both_ "faxq" and "hfaxd" have exited. >>> >> Why do you have the ignore_error? That's preventing the behavior you >> want, >> >> startd/ignore_error The ignore_error property, if set, >> specifies a comma-separated list of >> ignored events. Legitimate string >> values in that list are core and sig- >> nal. The default is to restart on all >> errors. >> > > One reason could be the same as I mentioned in the long-running thread > last week: if you don't set ignore_error, then any failure in an > external script you may invoke -- such as PCL2FaxCmd dropping core -- > will cause the whole service to get restarted rather than just failing > the current operation. >
I just can't help but think that a more extensible service check, like traditional HA agents, is where we are headed with this... -- richard
