02-03-2015 04:02 AM
Hello,
can somebody help please
*******************************
Enviroment:
I have a storage unit group ( mystug) including two STUs which is basically two media servers meida_A and media_B
mypolicy is created with one client running as cluster node_A and node_B
node_A can communcate with the master and media_A ONLY
node_B can communcate with the master and media_B ONLY
i setup mystug as Failover ( choose the first storage unit in the list that is not down or out of media )
****************************************
The issue:
the backup for mypolicy will be succesfull if the cluster active node use the corresponding media otherwise it will fail with communiation error.
*************************************
The question:
is there a way to run mypolicy backup using mystug and check if a communication exist between the media and client or use to the other stu ? failover option is not a solotion for my case
*****************************************
Thaanks
02-03-2015 12:12 PM
Sorry no.
You need to split that policy up in two - each policy with its own storage unit.
That does not make it a storage unit group, but that is the price you pay so to say, for node_B not being able to communicate to media_A
Any reason why node_B can't communicate with media_A ?
02-03-2015 08:32 PM
The question:
is there a way to run mypolicy backup using mystug and check if a communication exist between the media and client or use to the other stu ?
The answer:
Yes, but you'll need to script it. You could run a script on your master that say does an nslookup of the VIP of the cluster. If its in location A, it activates Policy A (de-activates policy B), if its in location B, it activates policy B (de-activates policy A).
But as Nicolia said, you still need two policies.
02-03-2015 09:01 PM
Thank you Nicolai and Riaan for your informative replies
mypolicy is for database backup, it cannot be two policies
Any reason why node_B can't communicate with media_A ?
cuz node_b and meida_B are in one private network and the same for node_A and media_A
iits usaull to face connectivy failures between media and client , so I think STU group should have more options to accommodate such scenario
02-03-2015 09:24 PM
02-03-2015 11:02 PM
Thanks Marianne for your suggestion
Im using media_A and media_B as dedicated media servers which are used only as media servers , as I understand from the Guide you provided , creating application cluster is NOT applicable for me. its only for non-dedicated media servers that run other application beside backup media software
02-04-2015 12:31 AM
Sara - I would go for a network reconfiguration to fix the issue. Remember a media server can have multiple network interfaces. Any workaround will be "bad design" and cause every day troubles (I hate those).
02-04-2015 01:06 AM
Nicolai, its not an issue in design
each media server has a multiple clients that they share a private network with use they are physically in the same box
so the idea each box has its own media server that used to backup clients in the same box for faster backups
the clustered servers is distributed into two boxes
02-04-2015 01:49 AM
I've seen this type of NetBackup configuration challenge when running stretch clusters. Each site as media server(s) but you obviously don't want the data to traverse the network. As explained by everyone above though, there is no automatic fix for this.
02-04-2015 03:29 AM
I wish I can find a permeant solution for my issue from Symantec
02-10-2015 01:20 AM
Is there a way ( to be scripted) to change the storage unit if a certain policy backup fails ?
02-10-2015 01:37 AM
If clients on box A can ONLY communicate with media_A there is no workaround or solution if media_A goes down or has a problem.
Nothing that Symantec can do about it.
Failover option in STUG will make backup go to another media server if problem is with media server.
If you want media_B to backup clients on server A if media_A has a problem, you need to enable public network access.
Maybe it is us not understanding properly...
If you have a diagram of server config with clear private and public network config and how hostnames have been configured in NBU, then maybe we can have a better understanding...