[Gllug] Advanced versions of cron for scheduling

[John Hearns]

On Sat, 2005-07-16 at 09:55 +0100, Richard Jones wrote:

> 
> Probably not very clear from my original email.  These jobs are user
> programs which must be run at particular times.  The programs are used
> mainly to fetch stats from remote SOAP services and change bidding
> prices based on these stats.  Since the remote services are
> failure-prone, we need to do things like handling jobs if they get
> "stuck" accessing the services, running bidding jobs only if stats
> jobs all succeed, not trying to run two bidding programs at the same
> time, and so on.
> 
> It all needs to run from a script of some sort (like a crontab), but
> update us by email, particularly when things go wrong.
> 
> > I'm not being sarcastic. Have you thought about using a scheduler/load
> > balancer like Gridengine?
> > 
> > http://gridengine.sunsource.net
> 
> I'm guessing this is for running jobs over clusters?  The website is a
> rather opaque, but I'll download it and have a look.

Well, for clusters yes, but load balancing/queuing systems were (are)
used on big systems and SMP machines which aren't clusters.

In the Gridengine jargon, you can have a master host and an execution
host on the same machine.
Gridengine has job dependencies, which is what you want, and will email
you when a job begins, ends, aborts or is suspended.
You can have epilog and prolog scripts run on each job queue.
You stop two jobs executing at the same time by setting SLOTS=1 in the
queue.

I don't think Gridengine is an exact fit for what you want, but give it
a try.








-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug