[ClusterLabs] [ClusterLabs Developers] Resource Agent language discussion

[Jehan-Guillaume de Rorthais]

On Tue, 11 Aug 2015 11:15:47 +0200
"Fabio M. Di Nitto" <fabbione at fabbione.net> wrote:
[...]
> >> In most systems, all commands required to execute a RA in shell are
> >> already cached in ram and requirements to re-run them are minimal (and
> >> could save a system).
> >>
> >> with Perl, there was no caching that I could see (even executing the
> >> command several times), with lots of I/O to load modules from disks.
> > 
> > If hitting 10MB on memory or disk is a problem on your server, your RA is
> > probably not your main problem by this time.
> 
> That is a bad assumption that shouldn´t be done either directions.
> 
> We could argue the other way around :)
> 
> If loading 10MB of $fancy_language_script fails to stop a service that
> is driving the server bad, vs loading 500k of shell that saves the node
> to be fenced, then the language become a problem.

Well, how much time a bash RA is buying you before the system eat the last
9.5MB compared to a $higher_language RA? The only situation you are safer is if
your service stopped eating memory when hitting 98% of it (and your cache where
your bash binaries were).

But in this situation, the OOM killer is probably already lurking around,
preparing to kill your service, calling its stonithd friend on the other node
to join the party right after.