cancel
Showing results for 
Search instead for 
Did you mean: 

Controlling Replication Jobs

John_Snodgrass
Level 4
Partner Accredited

I'm trying to figure out how I can limit the number of replication jobs that will run at one time.  I'd like to make it so only 1 job is active at a time.  This is more to control the data going out the replication network interface then controlling load on the disk pool.  I have a pretty good feel for doing that with the concurrent writes in the STU and Limit I/O streams in the disk pool, and using an SLP window to force the replication to run outside of the backups.

The WAN connection we have is very slow, and it's not going to be changed, already tried to fight that battle.  I've already set the Max size of the job to 100GB, and max images to 50 to try and keep the jobs relatively small.  The disk resource multiplier is set to 2, not sure if that will do what I want anyway.  I'm open to any suggestions.

The NIC on the media server is 100Mb/s, and the WAN is 50Mb/s

NetBackup 7.6.1.2

Redhat Linux 6.7

Single server doing everyting.  Getting more memory added soon. (80GB, --> 128GB)

3 REPLIES 3

John_Snodgrass
Level 4
Partner Accredited

Can you control the window close behavior?  I did see something posted awhile ago asking for input on what people wanted to have it do.  I have not seen anyting that says how it can be controlled.  If I can make the jobs stop at the end of the window, I'm good with that as well.

05/18/2016 15:00:00 - Info nbreplicate (pid=28417) SLP window closed at 05/18/2016 15:00:00, window close behavior: Continue processing the current image

PatS729
Level 5

Hope you are grouping images processed by a SLP that ways you can limit the jobs. If you have multiple SLP's then it would not help as all of them would try to start replication/duplication, but then you can define the window accordingly.

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Is your DR only used for accepting backups from Prod or is it active? If not active you can try to put the max jobs on the STU in DR to 1. When working with duplications and needing to run less streams I create a second storage unit for duplications only, i set this stu to have a lower concurrent jobs setting. With replication we obiviously don't specify a storage unit anywhere but maybe it still works on the same premise in the background.