On Thursday 25 March 2010 04:39:34 Morty Abzug wrote:
> I'm resurrecting an old thread about a patch I submitted some time
> ago. The idea was to implement "poll" (or "empty") jobs that do
> nothing but check backup intervals, and then implement a
> "MaxIncrInterval" analagous to "MaxDiffInterval" and
> "MaxFullInterval".
>
> Reasons to use this:
>
> * Better behavior for clients that are down (i.e. laptops that are out
> of the office, desktops that are turned off, servers with hardware
> problems.)
>
> * Better behavior for bacula servers that are down during scheduled
> incremental backup. This way, the incrementals will run shortly
> after the server comes back online.
>
> * Achieve AMANDA-like scheduling with bacula. That is, if one sets
> Max{Incr,Diff,Full}Interval policies and poll/empty jobs, one could
> do away with manually scheduling incrementals, differentials, and
> full backups altogether.
>
> An earlier email about this:
>
>
> http://www.mail-archive.com/bacula-devel@lists.sourceforge.net/msg04339.htm
>l
>
> Kern seemed to like the concept, but objected to the implementation.
> In particular, Kern wanted an Admin job to do this. References:
>
>
> http://www.mail-archive.com/bacula-devel@lists.sourceforge.net/msg04711.htm
>l
> http://www.mail-archive.com/bacula-devel@lists.sourceforge.net/msg04340.htm
>l
> http://www.mail-archive.com/bacula-devel@lists.sourceforge.net/msg04723.htm
>l
>
> I then stopped caring about this because of "Reschedule On Error",
> which solved the original problem I was trying to solve, i.e. laptops
> that weren't available during the backup period.
>
> However, I'm now trying to take advantage of "Allow Duplicate Jobs =
> no". It wasn't working. Looking at the bacula-users list, this is
> apparently a known interaction with "Reschedule On Error". So I've
> disabled "Reschedule on Error" and I'm back to looking for a solution
> to laptops. And this seems to be it.
>
> So, Kern: while I understand why you want this feature generalized, I
> really think it makes sense to have the "poll"/"empty" job as a type
> of backup job rather than an admin job. In particular, users can then
> think of backups as having a simple promotion hierarchy, from
> poll/empty, to incr, to diff, and then full. Implementing this as a
> new type of admin job may make the bacula command set more sparse, but
> will also complicate the lives of anyone looking to comprehend the
> promotion hierarchy.
>
> Also, the code for this is working now, fairly trivial, and solves a
> real problem. The code for a new type of admin job hasn't even begun,
> does not seem trivial (at least to me), and I, personally, haven't yet
> seen anyone clamoring for it.
Would you send us your latest code based on the current development stream?
Best regards,
Kern
>
> - Morty
------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Bacula-devel mailing list
Bacula-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-devel
