Hi Lloyd, all,

Am 10.11.2026 um 16:50 schrieb Lloyd Brown via Bacula-users:
It's true that the majority of the VFs were never restored from, but some certainly were.

This really gets to the crux of my original question.  Why is the cycle bad?  Suboptimal performance, perhaps, though I'd argue that for disks the difference is negligible.

We stumbled over that because performance was so bad that users actually couldn't complete their copy and virtual backups in decent time, couldn't saturate their tape drives, and found that SD threads, doing such jobs, wasted more than 90% (iirc) of a CPUs time going through bootstrap records. Negligible only if you accepted a whole university's backups to simply not take place any more :-)

  But what makes it bad/incorrect in some way?  That's what I've been trying to understand.

Optimizing bootstrap record handling in the way it was implemented required removing the ability to revisit volumes once processed.

I'm pretty sure it could have been done differently.

I'm also pretty sure that balancing effort, impact and potential risk of new bugs was done and the result was what we have now.

You should find the comment

/*
 * ***FIXME***
 * Also for efficiency, once a bsr is done, it really should be
 *   delinked from the bsr chain.  This will avoid the above
 *   problem and make traversal of the bsr chain more efficient.
 *
 *   To be done ...
 */


in the match-bsr.c source file of the SD, which indicates the developers could imagine alternative solutions, as well as

#define new_match_all match_all
/* Code to test the new match_all code. Can be disabled with a setdebug command */
int use_new_match_all = 0;


which indicates you could probably even test the impact, although you'd probably have to dig out and modify the, by now, probably hard-coded DIR code that prevents the "loops" to be created in bsr files.


I would assume that Eric and team would be considering a robust implementation that raises the existing limitation. Some of our useres surely would appreciate that. There are really critical code paths involved, though.


Cheers,

Arno

--
Arno Lehmann

IT-Service Lehmann
Sandstr. 6, 49080 Osnabrück



_______________________________________________
Bacula-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/bacula-users

Reply via email to