03-26-2012 08:22 AM
Hi NBU Folks,
I have the following requirement for backup start and stop script. If you have any that matches this, appreciate if you can share.
Comments, Suggestions most welcome
There is asynchronous Shadow Image (SI) that runs every day during OFF peak hours. Since the asynchronous SI "Syncs" with the mirror only during Shadow Image process and then splits, the DELTA change is quite much and the Shadow image can runs for anywhere between 4-6 hours.
If we call the SI process through bpstart_notify what we noticed is, as soon as the backup job gets kicked in, NBU starts the bpstart_notify script and in parallel also allocates the Tape Drive resource. Now if the SI runs for 6 hours, it keeps hold of the allocated drive for hours hence the resources are not utilized well.
Requirement.
1: Is there a way we can remove this pre-allocation of drives? we are using multi streaming.
2: One option is to schedule the SI job through Cron and call the backup using bpbackp -i policy_name -s schedule_name, but then we need one cron for each schedule. there are 3 schedules in our environment and expecting 16 such SI, that would mean 48 crons which will be too difficult to manage.
3. We are looking to initiate the SI from cron and not use bpstart_notify so that tape resources are not wasted at the same time utilize the advantages of Netbackup policy scheduler.
we are already working on a solution and shall share soon if successfully implemented.
Hope to get some response.
Thanks!
Satkay Satish
03-26-2012 10:27 AM
Why not use NBU Snapshot Client (Enterprise Client license) to automate the process?
Please mention your NBU version and OS on the client where Shadow Image backup needs to be taken.
It is a lot easier to point to correct NBU manuals if we know the version.
Scripts are specific to OS. You mentioned 'cron' - this means 'some sort of' Unix or Linux, right?
See NBU 7.1 Snapshot Client Admin Guide http://www.symantec.com/docs/DOC3661 for instructions on how to automate backup for Hitachi_ShadowImage using this feature.
03-26-2012 04:45 PM
03-27-2012 05:54 AM
Hi Marianne ,
Thanks for your response. Snapshot will be a good option, but it requires Netbackup client on each of the primary host.
My setup is having a Storage Layer in between the client and actual Storage.
The Shadow Image is done at the Actual Storage Level. And then the volume is mounted by the Storage layer before presenting it to the NBU media server hence Snapshot client may not be feasible. Still I'll give a try at later stage. Thanks for the suggestion.
My setup details as below.
NBU - 6.5.5 ( I know old, we are upgrading soon to 7.1 or 7.5 if available )
Media server mounting the Shadow Image - Linux RHEL 5.3
Actual Storage - Hitachi VSP.
Thanks for the link in doc for " how to automate backup for Hitachi_ShadowImage ".
BTW: I'm thinking of the trick to enable the policy soon after SI completes and Disable it after backup
03-27-2012 05:56 AM
Hi Yasuhisa ,
You're correct. Its the Netbackup design which cannot be changed. Hence I'm thinking of calling the Shadow Image process through Cron, once this completes, enable the policy .
The policy will be disabled after the job completes.
Thanks!
03-27-2012 07:02 AM
Had a similar thread a week or so ago - my suggestion then was to have a dummy policy which backs up just one small file.
This has a bpend_notify.dummypolicy script that does everything required and then fires off the main policy using the bpbackup -i option
Perhaps this could help you?
03-27-2012 08:40 AM
Hi Mark,
Thanks for your response.
The option that you suggested is really fantastic. However it requires 2 policies per client.
What is your view if we just enable the policy after Shadow Image and disable it once it completes?
Thanks!
03-27-2012 08:54 AM
To be honest i would go with my suggestion, even though it means having an extra policy
The main policy doesnt need any scheduled windows (just the schedule itself) - as the first one kicks it off - so no need to disable the policy as it wont run without being fired off manually
03-27-2012 06:32 PM
Thanks Mark. I'll let you know the option we implement.
04-03-2012 01:29 AM
Since the Shadow Image is asynchronous where the data is synced only during SI, the data on secondary copy is important and we didn't wanted the control for SI to be with Netbackup
I've compared the two options as below.
Will be discussing with our architect and finalize the approach.
Sr# | Feature | Policy Disable Option | Dummy Policy Option | Comments |
1 | Triggering SI | Cron | bpend_notify script of dummy policy | Policy can be wrongly triggered by DC causing the SI to kickoff. Cron will help to keep control with SE Also will be helpful when no resync is required |
2 | Policy Count | Single Policy | 2 Policy for each SI backup | |
3 | Change in Netbackup | Required to enable/disable the policy | Policy enable/disable not required |
04-03-2012 01:31 AM
Hi ,
I also need some help for bpstart & bpend notify scripts. Does anybody has ready scripts for multistreaming jobs?
http://www.symantec.com/business/support/index?page=content&id=TECH69986
Thanks!
05-21-2012 02:31 AM
We have finalized to go with policy disable option.
Also to note that we are not utilizing the quick sync , split option in HDS.. If that used there will be no delay for sync and no such requirement...