On Thursday 06 October 2005 19:21, Ross Boylan wrote:
> On Thu, Oct 06, 2005 at 09:09:57AM +0200, Kern Sibbald wrote:
> > On Wednesday 05 October 2005 19:52, Ross Boylan wrote:
> > > On Wed, Oct 05, 2005 at 11:42:09AM +0100, Russell Howe wrote:
> > > > Ross Boylan wrote:
>
> ...
>
> > > 2) single jobs. Priorities may influence which of several
> > > simultaneously scheduled jobs start first, but this is not completely
> > > reliable.
> > >
> > > Am I following correctly?
> >
> > Yes, with the exception of item 2. The scheduler *should* normally start
> > simultaneous jobs in priority order, but I warned users, that if they
> > want to be 100% simply defer the start times appropriately. The
> > scheduler cannot possibly schedule all jobs to infinity, so it works on a
> > two hour basis, thus without a mathematical proof, I cannot be 100% sure
> > what will happen in all cases.
>
> I'm not sure how literally you meant "proof", but if you give me some
> premises I'm game for seeing what conclusions could be drawn. Or is
> the issue more that you're aware of some weird cases where things
> would fall down?
As pointed out, Bacula "pre-schedules" or examines jobs in advance in either
one or two hour periods (I forget which). It then sorts the jobs by start
time and priority and runs them at the appropriate time. I have not
thoroughly tested the end point conditions (i.e. what happens if one or more
jobs fall exactly on the end point of the time fram Bacula is considering).
Any more details, you would have to get directly from the code ...
>
> Ross
--
Best regards,
Kern
(">
/\
V_V
-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
