Re: depend take 2

[Jan-Henrik Haukeland]

Martin Pala <address@hidden> writes:

> Jan-Henrik Haukeland wrote:
> 
> >Hmm, the depend functionality is much better but still does not work
> >as expected. I'm going to work a bit with on it. In case anyone are
> >planning doing stuff with changing control.c let me know.
> >
> >
> 
> I looked on it too - i think it will be probably better to make it
> more independent on the rest of control.c. Probably it will be better
> to use model proposed by Jan for it (dependancy matrix) => build
> process dependency relationship outside process structure (avoid use
> of process state variables).

I just realized that I'm not going to be able to finish a release of
the zervlet system before after Christmas. So I can let it rest for
some days and work a bit on monit which I haven't done for a while. I
want to test and see if it is possible to refactor the depend logic a
bit before we do a beta release.

> In addition i think (again) that it will be better to use statement
> "depend foo" as declaration for "slave" processes that they depend on
> process foo, for example to define that bar depends on foo:
> 
> 
> new proposal:
> 
> ###
> check foo with pidfile "/var/run/foo.pid"
> ...
> 
> check bar with pidfile "/var/run/bar.pid"
> depend foo
> ...
> ###
> 
> instead of present:
> ###
> check foo with pidfile "/var/run/foo.pid"
> depend bar
> ...
> 
> check bar with pidfile "/var/run/bar.pid"

If you mean, foo->bar, that is, foo depends on bar to run before it
starts, I think that the present syntax is better, since you can sort
of read it like: "check foo and depend on bar to be running".

-- 
Jan-Henrik Haukeland