Generally speaking, you don’t want to be mucking around with the contents of your disk backup units except under extreme circumstances. In fact, I really recommend that you don’t do so unless you 100% know what you’re doing.
So this post is all about that 1-5% of the time where you may find it necessary to say, search for a saveset that’s reported in the media database that you’re having problems accessing from the disk backup unit.
It’s actually trivially easy, once you know how.
You may be familiar with the following style of query:
# mminfo -q "savetime>=24 hours ago" -r "volume,client,level,sumsize,savetime(22)"
The (22) at the end of the savetime report parameter tells mminfo to allow 22 characters for the reporting of the savetime. The benefit of this is that you get not only the savetime date, but also the time as well.
NetWorker actually allows you to put that (number) postfix onto any field that it can output in mminfo. This can output additional information, such as the above, or give more room to output longer fields, or even limit the size of fieldnames when you don’t need too much information. (E.g., if the first 4 characters of all client names can uniquely identify the client, you might limit the client to 4 characters in an mminfo report.)
Now, where we’re heading with this, is that the sorts of filenames used for the savesets written to disk backup units are not some random collection of strings – they’re actually the long saveset ID.
Consider then a filename of:
/d/nsr/02/63/05/cd3a182f-00000006-7b7801de-497801de-01871a00-3d2a4f4b
This isn’t just a random filename, it’s the saveset ID, but just in a format you may not used to.
To get the long saveset ID in mminfo output, we use the (number) postfix on the ssid field. This would be as follows:
# mminfo -q "ssid=071462366" -r "ssid(53)" cd3a182f-00000006-7b7801de-497801de-01871a00-3d2a4f4b
With that information in hand, you can then search for a file with the same name as the long saveset ID on disk.
You can also do a reverse lookup. Say for instance, you know there’s an issue with a particular saveset file on a disk backup unit. To find out what the actual saveset ID is for this saveset, you can run the counter-query:
mminfo -q "ssid=cd3a182f-00000006-7b7801de-497801de-01871a00-3d2a4f4b" -r ssid
So, there you go – very easy!
Hi!
Thanks for this usefull informations.
Is there a way to select more than one backup level through the mminfo util.
Something like mminfo -q “level = full, level = 5”?
I have to build some scripts, which uses mminfo to match the right tapes. But I didn’t find any way this time.
Unfortunately there’s no way to select multiple levels simultaneously. If I’m trying to optimise scripts in this scenario, I select all levels, and include the level in the report section; that way I can weed out the reports that I don’t want.
One of the quirks of NetWorker is that repeatedly running a lot of media database queries or operations in rapid succession can swamp daemons and return unexpected results. While this has been massively improved in successive releases of NetWorker, I still try to avoid multiple queries in short succession, and thus it’s better to return more information and simply weed out what you don’t want rather than doing several queries in a row.