cancel
Showing results for 
Search instead for 
Did you mean: 

the current master server is limited to a maximum of 256 drives

Kai_8z
Level 3

we try to define new tape drives. Running the device configuration wizzard brings the error Message:

the current master server is limited to a maximum of 256 drives

my problem is that we have 200 drives and want to add 8 new... 

so why he said 256?  and is there a possibility to change this maximum number?

 

 

8 REPLIES 8

pats_729
Level 6
Employee
I have not heard of any limitations on NetBackup but I guess SG drivers on Linux may restrict it as it supports device ids from /dev/sg0 - - /dev/sg254.

Do you have all 200 tape drives connected to one system or multiple systems involved? Please put some focus on it as it will be very interesting to know.

You can try applying resolution from article herewith https://www.veritas.com/support/en_US/article.100029863

i also never heard this.
this is a clustered master. Both server has 200 Devices, and behind we have 8 Media Server.  But each node has it own devices.
what is also strange, that this message with the 256 appears before i can mark all already defined drives...  so after approx 160 drives marked to use it comes up with the message the not more then 256 are allowed...

Nicolai
Moderator
Moderator
Partner    VIP   

Calling for assistance from @mph999 

Hi @Kai_8z 

In my opinion, I think this whole question is pointless - just because it is possible doesn't mean it should be done. 

Rather than trying to figure out how to add more drives - why are you attempting to use so many in the first place. A single server will be unable to adequately stream to more than handful of drives (maybe 16 if the server is suitable configured and resourced) and the input side of the equation of getting the data into the server so it can write is something else again. 

I suggest you review what you are trying to achieve - I struggle to see a situation where a single server would need so many tape drives attached (especially on a clustered master server) - does the master really need so many drives? The whole point of having media servers is to scale out the load.

Just my 2 cents.
David

Hi David.

the question was, why NBU said we use more then 256 Drives but we only use 200.

i hoped someone here had this problem before and we will have a fast solution
i also opened a ticket at veritas, but this take time we don't have.

We do this NOT because it is possible, we do this because we have the need, and believe me it works good and fast!

1. we use only virtual drives from a Virtual tape Lib with real Tapes at the backend
2. the most of the 200 Drives on the master are NDMP drives and so the "workload" is on the filer and not the master.
and believe me they are fast.

the other "normal" drives are mostly placed on 6 media Server we talking here from about 300

the most people think if you have 500 Drives all of them must be used the whole day parallel in best performance
but the reason for so much drives is that a lot of clients are small and slow, and will block drives for hours.
to be virtual make the work with a mass of clients really easier as  with real drives.
And i don't want to discuss why we don't backup to disk.
we Want Backup on Tape for security reasons. and at the end of the VTL we have Real Tapes. it is fully automatic and work good for years

we are talking from about 16000 Backup Jobs a day only on this Master / Mediaserver combination.
and we have a view of this constructs..

and again it works better then ever before with real tapes.

Kai

 

 

 

 

 

Hi Kai

While I disagree with your reasoning here's some thing you can check. 

As has been said by others, I believe you will find that the Linux sg driver (which is used by NetBackup for connecting to tape drives) has a limit of 256 devices. The sg driver is also used by other scsi type devices in the operating system such as disk drives and robot control.

You can check to see how many are being used by the OS by looking in the /sys/class/scsi_generic directory. To determine what each device is being used for is more complex, but you can run the following shell command  (single line) as root to get an indication of the type of device each is using (aka the model) :
# for d in `ls /sys/class/scsi_generic/`; do echo -n "$d: "; cat /sys/class/scsi_generic/$d/device/model; done

If the number of sg devices is 256 or close to it, this may be your problem - is there anything in there that can be removed (unlikely but possible) from the system. 

I am reasonably sure the NetBackup limit is fixed at 256 drives per host (others may be able to confirm). The reference to the bpconfig command supplied by others is only to reduce the 256 limit, not expand and this option has been depreciated anyway.

As for not using disk based backups - the scenario you describe is one of the reasons why this should be a good solution. Not sure what you have against this, as I would have thought that managing a vastly smaller number of tape drives and tapes would be simpler than what you have described. 

Finally, although I cannot answer your actual question - support should be able to provide an answer, can you not confogure additional tape drives on your other media servers, if this is such an urgent problem for you?

Good luck
David

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

I am always wondering why VTL is at all being used.
IMHO - the technology is so 'last century'...

VTL made sense when backup products did not properly understand how to use and manage disk and disk vendors then invented a clever way to make their disk look like tape.

Disk vendors and NBU made enourmous changes to their products since NBU6.5 days.
With OST plugins, NBU and OST vendors have been able to provide better performance than tape.
Most OST vendors also support FC transport.

Maybe a good time to consider OST?

You can check the HCL to check support for your vendor.
Master Compatibility List: http://www.veritas.com/docs/000033647

X2
Moderator
Moderator
   VIP   

@Kai_8z wrote:

we are talking from about 16000 Backup Jobs a day only on this Master / Mediaserver combination.
and we have a view of this constructs..

and again it works better then ever before with real tapes.


I was reading some documentation today and got curious about the number of jobs on one of our backup domains. I consider out backup infrastructure (2 prod, 1 test) to be small one. The busy backup domain (1master + 2 media) does 7500 jobs/day. How many LTO7 tape drives do we use for it? Eight and they are not busy all the time. Reason being that we have a storage array using OST which then replicates to a similar one. After that tape copies of backups kept longer than 5 weeks are made. All these are done using SLP.

8 media servers! How do you manage downtime when upgrading? I need about 5-6hrs downtime for 4 servers in a domain during upgrade.