Hi, śr., 8 kwi 2026 o 23:00 Lloyd Brown <[email protected]> napisał(a):
> Hey, all. I'm in the middle of designing the next generation of our > backup system, which will likely be based on Bacula 15. This time, > we'll be using both disk and tape, vs the disk-only we've run before, > which brings up a question or two that I haven't encountered before. > > When running VirtualFull jobs, does the source material (previous Full, > latest Differential, subsequent Incrementals), need to all reside in the > same Pool? No. There are other limitations. > In multiple pools but available to the same SD? > Good question. I would say you need all at the single SD to work. But it is a conservative approach. I have not tested VF on multiple SDs yet. > For this system, we're planning on 2 SD hosts, with local disks, and a > tape library attached to one of them. We thought that we could keep the > Fulls/VFs entirely on tape, only store the incrementals on disk, and > write new VFs directly to tape. Good choice! > As a result, we trimmed down the size > of our disk pool for cost reasons (HDDs and SSD are both crazy expensive > right now). I don't think I'll have enough space in the disk pool to > store Incrementals *and* a previous Full/VF. If that's true, I need to > re-think my design a bit. It may mean we just do true Fulls to tape, > and Incrementals on disk, and not get to use VFs at all. > No, not needed. You can have your fulls saved on your tapes and incremental on disks. One caveat, you have to setup your Full Pool to close (status in ['full', 'used']) before you start to write the next Virtual Full. This is to prohibit read/write for the same tape during the Virtual Full Job. This means your previous (V)Full needs to be on the separate (closed) tape then the one you write to (status=append). best regards -- Radosław Korzeniewski [email protected]
_______________________________________________ Bacula-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/bacula-users
