NBU device monitor does not show tapes in drives

[Somniumus]

Master server = clustered two node RedHat Linux, NBU 7.1.0.3 installed

Two Win2008 R2 media servers, MSDP used.

Tape robot is a MSL8096 with 3 drives installed.

More then often we have one, two or sometimes all drives represented with a green arrow up and red arrow down icon. This sometimes occurs after failing backups with error codes 2, 252 and 2009 and error message "tperr load operation" from activity monitor.

After checking with robtest s d command we notice the drive(s) still have a tape in them (see example below)

drive 1 (addr 1) access = 1 Contains Cartridge = no
drive 2 (addr 2) access = 1 Contains Cartridge = yes
Barcode = ALM625L5
drive 3 (addr 3) access = 1 Contains Cartridge = no
 

Weirdest thing is that the Netbackup device monitor shows nothing in drives. According to the Netbackup databases these were robot = "NONE" - i.e. Netbackup believes these 3 tapes to be standalone.

Moving the tape with robtest m d2 s<slotnumber> to an empty slot in the robot and running the inventory from the NBU console seems to solve the problem, but this issue seems to repeat itself after a day or so.

Anyone encountered this problem before, and how can it be resolve permanently??

[Mick_Scott]

A very common problem with NBU is if  a tape is in a drive and new media is loded into the library. If the new media happen to use the tape slot used by the tape in the drive then NBU can't unload tape from drive and downs the drive.

[Somniumus]

In our case new tapes were put in the library in the morning, NBU ran its inventory and during the jobs ran in the evening and overnight the problem occured.

I'm still confused on how and why this is happening because it does impact vault duplications and catalog backups.

[Andy_Welburn]

In our case new tapes were put in the library in the morning

How do you insert them into the library?

If they are manually put into any available slot then, as Mick says, if these slots are actually 'occupied' by media that is currently in drives performing a backup then you can have problems as these loaded media have no slot to be returned to.

[Anonymous · Accepted solution]

I am having a similar experience of late, so to track down a pattern have decided to implement a drivewatch script in cron.

Check on every hour if there is a drive DOWN, and email me. Also show me if there are tapes in the robot.

NetBackup will not show the tape in the GUI anymore as I think it has possibly attempted the unload and completed that handover.

Your practices are sound.

No matter if you use the 12 mailslot or 3 mailslot MAP or no slot config of your MSL8096. As soon as you interact with the robot by removing or adding media into slots outside of NetBackup - NetBackup has to know about it ASAP via running inventory of your library in NetBackup.

I am using the vm.conf variable AUTO_UPDATE_ROBOT

So I have to be aware of the scenario descibed below.

you could have operations rerunning failed backup jobs from the night before during the day and while tapes being returned into the library, but not via the MAP

Recommend to use the mailslots (inport/output) when you can.

But as Mick has said. Beware, as if you open all the media cartridges and place whole batch of tapes in AND inventory the robot while a backup job is runnning. This tape in the drive has a source slot where it came from. It needs to return there. This is when I have seen my drives go down. When the slot has become filled by another tape. Tape in drive has nowhere to go and NetBackup downs the drive.