I have discovered this problem with 7.3, 7.4, 7.5.2 and still present in 18.104.22.168.
My env is a quite simple one, all lan based clients (windows and solaris), saving to
datazone server on Windows 2008R2, no storage nodes. As devices, afew adv_files and
an overlan NEO Library with LTO5 FC devices.
Let's demonstrate the problem :
0) Elect an advfile device for the test. Advfile is made of 2 devices, a writable .WO and a readable .RO part (to restore/clone/stage from)
1) Have afew backups written to selected AdvFile device, enough data to see the problem.
2) prepare the staging of the advfile device : mminfo then nsrstage SSIDs ....
3) launch nsrstage that will read from .RO part of the advfile device you elected.
4) DURING NSR STAGE OPERATION, start a new backup to the advfile device. New backup will use the writable part of the advfile.
5) ## IF ## the nsrstage is finished before activity on the writeable part of the advfile is done (save from part 4- in this sample), the record of the mediadb assosciated to the advfile writable part volume's name has the volume capacity NOT updated with capacity difference reclaimed by nsrstage.
Rewording, During Staging operation, busy-writing AdvFiles have used capacity info from media database not updated (decreased by) with reclaimed volume space.