Re: starting service order

[Jan-Henrik Haukeland]

On Jul 7, 2004, at 6:23 PM, Christian Hopp wrote:
> (1) I think we really need to dive deep into graph theory then for a
> relialble implementation.  (definitely not my field of research) (-:

Shouldn't be to hard. Implementing dependecies was the hard part, 
implementing a "use" functionality simply means to shuffle the 
servicelist so services are started in the prefeered order. Okay, maybe 
some graph theory but not much. I'll put myself up as responsible for 
this function since I like to fool around with lists/graphs and 
pointers =8-o

> (2) Interesting but too dangerous.  Breaks the concept of having a
> validation loop.  Then we would have signal based tests and a 
> validation
> loop.  The signal based code might break with the multitreaded code.  
> And
> it breaks with the "unix" idea one small tool for each task.  I think 
> init
> is there where its supposed to be.  I think we would not get past a 
> proof
> of concept and wouldn't achieve anything a user would adept.  Just my 
> two
> cents.

a) I agree with your concerns, this is not trivial stuff. However, I 
suggest we put it on our TODO list 
(http://www.tildeslash.com/monit/next.html) but with a low priority. If 
we managed to implement this (or maybe as a better hack than using 
signals) it would be a very good feature, so monit does not need to 
depend on pid-files only.

b) init is absolutely where it's supposed to be. monit will not replace 
init, but can replace execution of SYS V rc startup scripts. The setup 
will look something like: init -> monit -> start-system-daemons. See 
Igor's general setup (which is not far from a sysinit replacement) here 
http://people.altlinux.ru/~homyakov/packages/monit/monitrc.d/

Cheers
--
Jan-Henrik Haukeland