Jukebox Barcode Question

[esserc]

I'm using 6.1 on Sloaris 8 on a E6500. I have a Sun L1100 (ATL a3000) jukebox and a Sun L700 (STK L700) connected to it. I have barcode readers on both and they are enabled with barcodes as labels, and that works fine.

I unload and load about 100 tapes into both jukeboxes daily. When I load them I do an inventory,
nsrjb -j <jb> -I -S <slot>.

Now, the jukebox knows the barcode. I can verify this by reading the element status. Networker knows the barcode because it can label blank tapes correctly, but Networker still loads each tape into a drive and reads the label. This takes us about 4 hours a day to do.

Does anybody know why Networker does this, and is there a way to change the behavior? Is it a bug? I believe 5.x worked correctly with the barcode, but I'm not positive.

 

[Matts]

If I'm not mistaken, NetWorker reads the barcode from the library. If it can match the label to an existing index then it's satified. But if it's a blank tape or it has never been used by the server, i.e. there's no index for it, it will load the tape into a drive and search for a label on the tape. It sounds like you're using new tapes...?

Good luck!

 

[TDun]

As long as the volume is already known to Legato it shouldn't prompt so, the question is, are you using the
correct commands.
To unload a volume, you should be using the nsrjb -x -T barcodelabel to remove it and the nsrjb -a -T barcodelabel to add it?
Another question would be, why are you unloading/reloading that many tapes every night?

 

[esserc]

To answer the first response: T
hey are not new tapes and the barcodes match the labels and are in the media index.

The second response:
The -x and -a switches are for remotely managed jukeboxes such as silos. I'm using localy attached jukeboxes.

 

[TDun]

So, what commands are you using to unload and load each tape and why are you doing so?

 

[esserc]

It's our policy to send every tape offside each morning. We do have a few that we clone and send off, but those are only the ones that we do a lot of restores from (eg user directories and mail).
Here are the commands I use:

The tapes get unloaded (If mounted in a drive):
nsrjb -u <volume>

Then sent to the mailslot:
nsrjb -w -j <jukebox> -P <port> <volume>

Once they are expired and returned they are deposited:
nsrjb -d -j <jukebox> -P <port> -S <slot>

Then Inventoried:
nsrjb -I -q -j <jukebox> -S <slot>

 

[TDun]

That looks ok as far as it goes, but -I can be slow as it will load tapes when it hits what it does not expect.
Does Networker know the barcode before you do the inventory? If not, then the volume must be read to identify the media label and this will slow things down considerably.

You are best off allocating a large timeslot to label a bunch of tapes, then unload/withdraw them (try about 30-40 at a time).
You can then deposit these and inventory as you need and it should be faster.

 

[tonyletigre]

hey instead of doing an inventory....I just relabel the tapes that I put in the jukebox..no inventory is needed

 

[esserc]

TDun:
The tapes are already labled, and the jukebox knows the barcode as does Networker. That's the problem. It seems to be a bug in 6.x beacuse 5.x worked as advertised in the manual.

Here's an interesting observation:

If you withdraw a tape into the load port and deposit it back without opening the load port door, the tape is barcode is read to do the inventory.

But, if you open the load port door before depositing the tape back in, the tape will be loaded into a drive to be inventoried.

This makes me think that is has something to do with Networker improperly reading the element status from the jukebox.

Tonyletigre:
Relabeling the tapes takes far longer than reading he lables, it throws off the tape usage counts (# of times recycled) and it keeps networker from trying to use all the tapes equally so you wear out some tapes faster than others. In other words, it's not a very good practice.