Re: additional feature for monit-3.0 (for clusters)

[Martin Pala]

Christian Hopp wrote:

> On Mon, 28 Oct 2002, Jan-Henrik Haukeland wrote:
>
>  
>
> > > Agree - some suggestions?
> > >      
> > A though one, a new name :-) If we do not find a good name (I'm
> > blank), we could combine, that is, use automonitor and autostart and
> > set the value properly in the parser, i.e. override the value of
> > autostart if it's true. (and document this behavior)
> >
> > BTW, using monit together with heartbeat is interesting, do you
> > think you could write a FAQ or man file entry for this Oliver? (When
> > we figure out what the statement should be)
> >    
>
> I am kinda confused about all that... maybe it's simply to high level
> for me... no peeking and poking the system.  So could some one give me
> a short reason/explanation... why we should do all these many
> confusing "set auto.*" things.
>
> Couldn't it be integrated in the set autostart?  What about
> "set autostart=[yes|no|manual]"?  As Hauk already mentioned... we
> shouldn't bloat the language.  With that we won't even have the
> problem with the "xor" thing.
>
> Christian
>
>  
Not bad, but i thing it is not clear enough. If we still would change 
the name for 'autostart' statement, the only name that occures in my 
mind is maybe:

'mode [ active | passive | manual ] '

where 'mode' is statement keyword (replacement for 'autostart') and:

active ... in this mode monit will watch the process and will restart it 
if it doesn't exist (equals present autostart=yes)
passive ... monit will start monitoring process but it doesn't do 
restart if it doesn't exist ((equals present autostart=no)
manual ... monit will enter 'active' mode after instructed to start this 
service, otherwise it woun't monitor this service at all

I'm not sure if it will solve this problem - maybe it will be better to 
keep 'autostart' as mentioned Christian. What about it?

Martin