Hello,

On 19.10.2005 17:00, William D. Hamblen wrote:

Thanks for the reply Arno,

Anyway, my question. I have the 8 loaded tapes assigned to two pools (6 in one and 2 in the other). I have run "update slots scan" and it all looks right and seems to be working (so far) but "list pools" shows 5 tapes in one pool and 1 tape in the other. Is this a bug? Something I have done wrong?


Oh, you will likely run in more trouble once you start unloading, loading, moving and especially updating tapes - 1.36.anything doesn't really support media from different pools in one autochanger.


Really!? It's not that I doubt you - I've been lurking on the list for awhile and making heavy use of the archives over the last week so I am well aware that you know as much about this as anyone - but could you be more specific about what might happen?

Sure. The problem is that Bacula doesn't reset the information in the catalog about loaded media in that case - it only restes information for volumes in the same pools as what is currently updated. Example: You've got a tape from PoolA in slot 1. You remove that and insert one from PoolB into slot one, and update the slots information. From then on, Bacula will have both tapes marked as being in slot 1 in the autochanger. Obviously, when it needs the tape from PoolA, it thinks it's available, loads from slot 1, finds the wrong tape, and is very unhappy about that.

(You could patch that with a one-line modification in the source, which is what I did, and it worked *for me*.)

Also, Bacula 1.36 has no concept of more than one autochanger, but that might not be a problem for you.

Thus far Bacula has sucessfully overflowed from one Full volume (tape) to the next and several times switched back and forth between tapes in the Full and Partial pools. I've even moved volumes between pools successfully - from the Default pool which I was using for testing to the Full/Partial pools (but I then relabeled those tapes if I remember the sequence correctly). It tracked all that perfectly. I only ran "update slots scan" to see what happened. So it seems to be working quite well. One thing I have not tested yet is running out of loaded tapes in a pool (maybe today, certainly tomorrow).

Then you might find the problems today or tomorrow. Brush up your SQL to fix the catalog information manually :-)

Anyway, I guess what I am trying to say is that if I hadn't just randomly tried "list pools" to see what the output looked like I would never have suspected there might be a problem.

Yeah, I suspect that that information is hardly ever used, only in case of automatic volume labeling. By the way, I found some major differences in the NumVols data and what the pools actually contained a while ago, and it had no bad side effects I noticed (not using automatic labeling, though).

Before investigating further, I suggest you give the development version 1.37.40 or .41 a try - it is still considered unstable, but works fine enough (here and in other places). With a straightforward setup like you have, I think there should be no unexpected problems. It implements many features around a more flexible volume management. But it does require a catalog database upgrade, and slight configuration changes.


Does this also require upgrading the clients? I'm a little reluctant to do that to 21 machines but if it's going to be necessary eventually I guess it's better now than when I have 40+ clients. :-)

Hmm. Difficult decision, as upgrading the clients will be necessary. Either way, once you want to switch to 1.38 (when it's available) another upgrade might be necessary. So, perhaps it's better to upgrade 21 machines now and later all of them than than waiting, and then upgrading 40+ machines twice :-)

I could probably get by with one pool for that matter. I separated into two because of the difference in retention times but I suspect the Partial pool is will be *so* much smaller than the Full pool that recycling tapes in the Partial pool will not be an issue.

Well, if one pool is enough for your needs now that might be the simplest solution.

Arno

 - Bill


--
IT-Service Lehmann                    [EMAIL PROTECTED]
Arno Lehmann                  http://www.its-lehmann.de


-------------------------------------------------------
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

Reply via email to