Feb 272010

While this is pertinent to all versions of NetWorker, it particularly seems relevant mentioning now, since as of 7.5.2, we’re now seeing revised messaging from NetWorker when a tape becomes prematurely full. These new messages now state:

nsrd media notice: LTO Ultrium-4 tape 800814L4 used 2039 MB of 800 GB capacity
nsrd media notice: NetWorker media: (Warning) 800814L4 marked full prematurely.
  Verify possible error on the device /dev/nst4, advertised capacity is 800 GB
  marked full at 2039 MB

Now, it’s worth noting here that normally if you get a tape fill up so soon that probably means there is an issue, and this version of the message, while only subtly different, is certainly more informative and that is a good thing. When we consider VTLs however, it’s a different story. In a virtual tape library, we normally want to use much smaller media sizes than the drive type we’re configured for. That way you’re writing virtual volumes that are 50GB or 100GB rather than 800GB. In my case referring to the above, my lab VTL uses virtual media sizes of 1GB (with compression).

So, how do you go about this? Well, it’s easiest to accomplish when you first setup the environment. You need to change the “Volume Default Capacity” of each virtual device to suit the allocated media sizes. To do this, in NMC turn on View->Diagnostic Mode, then when viewing device properties, enter the appropriate size in gigabytes (followed by “G” or “GB”) in the “Volume default capacity” field of the Configuration tab, shown below:

Changing default volume size

Now, if you can do that on your VTL devices before you start labelling volumes, you’re done and dusted. However, if you’ve previously labelled your media, you either have to relabel the currently blank virtual media or wait until NetWorker gets around to recycling the currently used media.

You can query mminfo to see what the default capacity is registered at – e.g.,

[root@tara ~]# mminfo -m
state volume                  written  (%)  expires     read mounts capacity
800801L4                2254 MB full 02/26/2011   0 KB     5    800 GB
800802L4                   0 KB   0%     undef    0 KB     5   1000 MB
800804L4                   0 KB   0%     undef    0 KB     5   1000 MB
800805L4                   0 KB   0%     undef    0 KB     3    800 GB

Now, what effect does this have to how much you can write to the volumes? The short answer is none. All you’re doing is adjusting the default capacity assigned to new volumes that are labelled in these (virtual) tape drives – and we can see what happens when NetWorker breaches the default volume capacity all the time in relation to physical tape – it just keeps writing until it hits end of physical tape. Nothing more, nothing less. So this means when you fill up your virtual media, NetWorker doesn’t complain at all:

nsrd media notice: LTO Ultrium-4 tape 800802L4 on /dev/nst3 is full
nsrd media notice: LTO Ultrium-4 tape 800802L4 used 2793 MB of 1000 MB capacity
nsrd media info: WORM capable for device /dev/nst3 has been set

Is this something you must to do? Well, no, not technically. However, remembering that I advocate a zero error policy, the above is something I’d definitely strongly recommend for virtual devices. Doing so will eliminate what would otherwise be false errors on the virtual tapes within the NetWorker daemon logs. That means if you have to search for media issues, or refer your daemon logs to your support provider for analysis, they won’t be seeing bunches of “tape filled prematurely” issues.

Sorry, the comment form is closed at this time.

%d bloggers like this: