Number of tape drives for concurrent backups

[spitman]

Good morning,

This isn't a question about multiple streams, etc...

I swear there is a global-type setting somewhere in NBU, that sets how many tape drives can be use at one time... i.e. I have ten tape drives in a library, and I know at one point that I set it so only 8 would be in use for backups at any one time (so I would have two free if I needed to do a restore)...

Bonus points--what is the command line way to view/change that setting?

Thanks in advance!

[Andy_Welburn · Accepted solution]

Can use the following to amend the 'concurrent jobs' (drives) for storage units:

bpsturep -label STU_name -cj number_concurrent_jobs 

& to check:

bpstulist -U

***EDIT**

For one STU you would obviously set this as -cj 8 to leave your 2 free in your scenario.

If you had more than one STU for whatever reason utilising your 10 drives, then your STUs would have to be limited so that they did not breach the 8 limit e.g. 1@ 5, 1@ 3 or 1@ 5, 1@ 2, 1@ 1 etc etc.

(Obviously the GUI is available if you want!)

[spitman]

Where is this setting in the GUI?

When I run the bpstulist, I only see an (8) if I run it with a -L (and I am just guessing, since I know the current setting, that that's what the 8 is for?)

> bpstulist -label rchnbump1-hcart2-robot-tld-2-rtxdxip01 -U -L

Label:                rchnbump1-hcart2-robot-tld-2-rtxdxip01
Storage Unit Type:    NDMP
Host Connection:      rchnbump1
Number of Drives:     10
On Demand Only:       no
Density:              hcart2 (14)
Robot Type/Number:    TLD (8) / 2
Max Fragment Size:    1048576
Max MPX/drive:        1
NDMP attach host:     rtxdxip01
 

is this (8)/2, does that mean 8 for backups and two for restores? Does "concurrent jobs" mean only ones kicked off automatically?

[Nicolai]

There is a global setting limiting the number of jobs per client - could that one be the culprit ?

GUI -> Netbackup Mangement -> Host Properties -> Master Server -> Global Attribiutes -> Maximum job per client.

There is not global setting of "Maximum number of concurent drives". It pureley a storage unit definition.

[Nicolai]

You will need bpstulist -label XXX -l (small -L) to see all of the settings

It don't look lige "maximum concurrent write drives" is show using -L or -U

[revarooo]

I am sure there used to be a setting to for this, but I cannot find it, so I am assuming it has been removed in later versions of NBU.

[Marianne]

The 'Number of concurrent drives' for this STU is 10:

Number of Drives:     10 

Meaning that this STU can use 10 tape drives simultaneaously.
If yo want to change this value to 8, this is the value you need to change with -cj option.

The meaning of this:

Robot Type/Number:    TLD (8) / 2 

Robot type is TLD - numeric value of TLD type is 8. 
Other robot types (such as ACS, TL8, etc) will have different numeric values.

The '2' is the robot number.

You probably have 3 or more robots in this environment:
TLD(0)
TLD(1)
TLD(2)

This STU is associated with TLD(2).

[Andy_Welburn]

The commands reference for 7.6 states the following for using the -L

    A long listing has these attributes for each Media Manager storage unit:
    - Label
    - Storage Unit Type (For example, Tape (0))
    - Host Connection
    - Concurrent Jobs
    - On Demand Only
    - Robot Type
    - Max Fragment Size
    - Max MPX/drive

Maybe it's different again for NDMP STUs .....

[spitman]

Andy, thanks for the screen shots... and I looked at the storage unit defined for the library, and it shows ten... I swear somewhere I'm still missing a setting that limits it to eight at a time. Is it maybe an SLP thing? The ones that are running most of the time are SLP copies to tape... 

[Marianne]

SLPs use Storage Units.

There may be values that were updated in SLP Parameters....

[spitman]

I haven't found anything in the SLP settings that would set this limit. There's still something I'm missing. This is driving me nuts!  :(  It's not maximum jobs, because the SLPs are all that are running, and the storage unit has a max of 60, so it's not hitting that limit...

[Marianne]

".. . storage unit has a max of 60... " Nope. The STU config that you posted above has concurrent drives set to 10 and MPX of 1. So, max possible jobs for this STU is 10. For SLPs, there is a 'multiplier' setting. This is a global value and not set per media server or STU.

[Pritesh_Pisal]

if you have backups which backed up while concurrent write drive were 10 then you cannot change the settings coz SLP holds that information with its version. 

if you do any modification in SLP it creates the version in the background

you can check that by following command

nbstlutil list –lifecycle <name>

e.g. if you made changes in STU concurrent write drive 8 and initiate the backup by using same SLP. it will backup with its another version. so you will be having 2 version of same SLP in background one with concurrent write drive 10 and concurrent write drive 8.

[spitman]

Sorry Marianne, when I said storage unit, I was referring to the disk storage unit that the SLP is copying to tape from.

[spitman]

Here's another thing that seems to be getting lost in translation: 

I have ten drives, and only eight are currently being written to at a time. I was looking to possibly change that to ten. So something is stopping it from using all ten drives at the same time. Does that make sense?

[spitman]

I may have been going down an unneccesary rabbit hole, and I apologize...

I started looking one by one at the jobs that were running and writing, and listed out the drives on a sheet of paper, and there are ten of them.

Then I started looking at the columns in the GUI. There are ten "External Media IDs" and only eight "Recorded Media IDs," which had been the column I was looking at. The External Media ID column was out of sight to the right.

The two that don't have Recorded Media IDs, I take it that is because they are tapes that have not been written to before?

[Nicolai]

Yes

[Marianne]

So..... to get back to your initial question....

Andy gave you the command to reduce number of max drives.

bpsturep -label rchnbump1-hcart2-robot-tld-2-rtxdxip01 -cj 8

I guess we can mark his post as solution, right?