Need some ideas -- Help!

[Nathan_Kippen]

I've been trying to figure out a way to ensure that everything that needs to get backed up on a server gets backed up....

I have several servers that I need to use multiple backup streams to my tape drives in order for each server to be backed up completely in the time frame that I have been given.  With that being said, I can't use the directive "ALL_LOCAL_DRIVES" because that only uses one stream, and NetBackup isn't smart enough to divide the data up into lets say 4 streams.

So I had an idea ... I specify all the current mount points (directories) that need to be backed up, throw in some NEW_STREAM directives, and then I create a new policy with an exclude list that excludes those same mount points that i backup in the original policy ...  and in this new policy I use "ALL_LOCAL_DRIVES" that way if the system administrator creates something that needs to be backed up, I don't have to worry about the new file(s) or directories not being backed up.

I had counted on getting an error 71 whenever this new policy ran because I thought the exclude list should have  excluded everything -- leaving nothing to be backed up.  However I'm getting Status 0s instead of 71s.  Which makes it harder to determine if any new data was found and backed up.

Does anybody have any ideas or a solution that are currently using for ensuring that everything is being backed up on a system without using "ALL_LOCAL_DRIVES" and without relying on the system administrator of the box?

Please ask questions if I need to clarify anything.  Thanks.

[Stumpr2]

---

Nathan Kippen wrote:

I can't use the directive "ALL_LOCAL_DRIVES" because that only uses one stream, and NetBackup isn't smart enough to divide the data up into lets say 4 streams.

---

ALL_LOCAL_DRIVES will work. It will divide up the mount points for you as separate streams. You should investigate why it isn't working and correct the problem.

What is Multistreaming and how is it turned on?

http://support.veritas.com/docs/267291

Message Edited by Bob Stump on 04-10-200711:10 AM

[Nathan_Kippen]

Bob,

Thanks for the reply, I was under the impression that it didn't work that way.

Do you know how NetBackup determines which mount points should go to which streams?  I have several servers with cold oracle databases that need to be backed up that are quite large.  Is NetBackup smart enough to not stick a group of large directories on the same stream?  Or Do I need to use an exclude list and exclude the larger(est) directories and build another policy to back those up?

Thanks,

[Stumpr2]

each mount point is a separate stream regardless of the amount of data on the mount point

from the NB 5.1 sysadmin guide vol1 page 135

Auto-discover streaming mode is invoked if NEW_STREAM is not the first line of the

backup selections list and the list contains either the ALL_LOCAL_DRIVES directive or

wildcards. In this mode, the backup selections list is sent to the client, which preprocesses

the list and splits the backup into streams as follows:

◆ If the backup selections list contains the ALL_LOCAL_DRIVES directive, NetBackup

backs up the entire client but splits each drive volume (Windows) or file system

(UNIX) into its own backup stream. See “ALL_LOCAL_DRIVES Directives” on

page 137

[Nathan_Kippen]

Bob,

Again, thanks for your sharing of knowledge and information. 

..points for you.

[Dennis_Strom]

If your setting up your own streams to insure the streams are balanced or to make sure that you have certain streams and yet do not have access to the systems you can run check_coverage.  Or, you can run bpcoverage -c clientname on each individual box and it will tell what is covered and what is not.  It also tells you what policy is covering the filesystem so you will know if it is being backed up twice.

[Stumpr2]

WARNING:

Do NOT use cross_mountpoints when you use ALL_LOCAL_DRIVES with multistreaming or else you will backup some data twice.