06-11-2014 12:31 AM
I read the post again and felt it is not clear :(
is it possible to make policy storage dynamic for the same policy? for example if the active node for a cluster is NODE_A use Media_A and if NODE_B is the active use MEDIA_B ?
Hello everybody,
I have two questions on virtual media server, if you could help please J
- Is it possible to use virtual ( or physical) media server ( tape backup) to backup virtual IBM cluster ( IBM power)
- How is it possible to use virtual media ( tape backup) to backup all virtual servers in one platform ( IBM)
I've read the Statement of Support for NetBackup 7.x in a Virtual Environment ( Virtualization Technologies :( http://www.symantec.com/docs/TECH127089
but it did not answer my concerns :(
Thanks
Solved! Go to Solution.
06-12-2014 12:52 AM
Hi
Still not clear on WHY you want to do this but you can do the following.
In your cluster (if it works like VCS and is a good cluster product :p) you should be able to have triggers (the ability to run scripts when the clustered application group starts up). You can use such triggers to do one of the following
06-11-2014 02:16 AM
Have you seen this TN?
06-11-2014 03:05 AM
yes, I read this TN
but my concern is how backup virtual cluster using one media server
in both platform that contanis both cluster nodes there will be one phyiscal media server
each cluster node must use the media server within the platform ! how is this achievable ?
in the policy there will one media server to use only?
06-11-2014 08:47 AM
This looks like something Ive been looking for myself and I dont know of a simple automatic way of achieving this. I work around it by using a policy per media server and then flip my databases (in my case) from one policy to the other. They dont often move about, so its not a big effort. Mine are service addresses , not virtual but the effect is the same I feel.
Jim
06-11-2014 09:04 PM
Sara I am not absolute clear on your qustion but based on my understanding I think appcluster is solution for such setup but I guess Netbackup virtual stirage unit does not support IBM cluster.
Let me know if I misunderstood and I will try to find answer for you.
Excerpts from HA Guide
Ina typical NetBackupenvironment,thestorageunitisassociated witha NetBackup entity, such as the master server or media server. In the configuration that is described in this topic, however, a storage unit is associated with the virtual name of the application in the cluster environment. The storage unit for that application is not associated with any specific node in the cluster and can failover to any node
in the cluster. This lack of association ensures that the storage unit follows along with the application, wherever the application is running in the cluster. The NetBackup master server uses the virtual name of the application as the media server name when it backs up the clustered application.
This configuration is supported for Veritas Cluster Server (VCS), Microsoft Cluster Server (MSCS), SunCluster , Service Guard, and HACMP.
DeepS
06-11-2014 09:47 PM
Hi,
Your questions is bit unclear.
Do you have one cluster node in on location and another in another location, and you want to back it up using the virtual name/ip to the respective site where it would be located at a specific time?
That would be what Jim touched on in the post above?
Then for your remaining questions
06-11-2014 10:26 PM
Thanks Marianne, Deeps , jim dalton and Riaan Badenhorst for your valuable input
Sorry for making this post unclear
But the real concern to back up a cluster using two separated media servers
I want to use one policy for the cluster with two media servs, ( if node A is active use mediaA, vice versa)
As jim Dalton and I are looking for automated way for achieving this ? either from GUI and having some script to do that?
06-12-2014 12:12 AM
What if node A is active but Media server A is down ? Do you want backup to fail ? what you can try. Create a Storage Unit Groups and add storage unit from each of the media server you want to use to this STU groups .Configure the policy with this STU Group.
DeepS
06-12-2014 12:32 AM
As I understand the question a storage unit group (as Depps mention) with fail-over as selection criteria sound like the right solution. It may not always be the optimal solution to prevent cross data center backup, but it will get you going and let the NBU admin review the incident in normal working time.
http://www.symantec.com/docs/HOWTO34499
06-12-2014 12:52 AM
Hi
Still not clear on WHY you want to do this but you can do the following.
In your cluster (if it works like VCS and is a good cluster product :p) you should be able to have triggers (the ability to run scripts when the clustered application group starts up). You can use such triggers to do one of the following
06-12-2014 03:34 AM
I think the WHY is for performance if its like my situation: servers A and B are media servers. If the resource is on A then it needs to use media server A (directly attached storage), otherwise the data will be going over the network to B.
In fact I have stunit groups: A then B , B then A.
Riaans idea of a trigger on failover is worthy of investigation if you want to automate.
Jim
06-12-2014 09:07 PM
jim for this performance issue we have virtual storage unit (app cluster) so that backup go to storage unit local to active node.But Appcluster does not support IBM cluster.
Excerpts from HA Guide
Ina typical NetBackupenvironment,thestorageunitisassociated witha NetBackup entity, such as the master server or media server. In the configuration that is described in this topic, however, a storage unit is associated with the virtual name of the application in the cluster environment. The storage unit for that application is not associated with any specific node in the cluster and can failover to any node
in the cluster. This lack of association ensures that the storage unit follows along with the application, wherever the application is running in the cluster. The NetBackup master server uses the virtual name of the application as the media server name when it backs up the clustered application.
This configuration is supported for Veritas Cluster Server (VCS), Microsoft Cluster Server (MSCS), SunCluster , Service Guard, and HACMP.
DeepS
06-13-2014 03:29 AM
Still not 100% clear on the requirement.
My guess is that nodeA and nodeB are clients, and that MediaA and MediaB are media servers at site A and site B.
I have never been able to find an easy solution to this question.
With some database extentions, it is possible to specify the schedule in the backup script.
Here each node will specify a different schedule in the script.
The policy will have 2 Application Backup schedules - each with 'override STU' selected.
So, nodeA script will specify the schedule name with MediaA STU,
nodeB script will specify the schedule name with MediaB STU.
We also had a customer who started the backup on each client using cron.
The script would first check if the cluster is active on the local node.
If so, start the backup (either bpbackup or db commands) and specify the policy/schedule for the local node.
06-18-2014 11:31 PM
Sara, Think a way to achieve this might be to have an alias Media_C and have NodeA point the alias to Media_A and NodeB point the alias to Media_B through the hosts file and use the Media_C in the policy. It is not pretty and will complicate troubleshooting of failed backups/restores.
It has been a while since I have done things like this in NetBackup, so is unsure if it is possible in the newer versions