Solved

SQL Cluster with multiple instances


Badge +2

We have a two-node SQL-Cluster with 10 SQL-Instances, load-balanced over the nodes.
Currently we fail to set up the backup. Only the instances running on node A are passing the backup successful. Instances on node B are failing with the following error:
[30:332] Description: Unable to get the SQL Version for the server

  • installed and check cv-software von slqcn1 and sqlcn2
  • added a windows-cluster in commvault and added sqlcn1 an sqlcn2 to this cluster
  • added all sql-instances to this windows cluster

Is this correct or do we have to create one cluster for each sql-instance in cv-software?

icon

Best answer by Mike H 29 March 2021, 17:45

View original

11 replies

Userlevel 2
Badge +11

YES, finally it is working!

As you described, we had to create a cluster group for every single SQL-Instance.

Thanks for your great support.


Hello @Sven Roth !

I guess these was the steps that you made for your SQL Cluster configuration:
(after 2:20)
 

Best regards,
Nikos

Badge +2

YES, finally it is working!

As you described, we had to create a cluster group for every single SQL-Instance.

Thanks for your great support.

Userlevel 6
Badge +13

Each clustered object created in commvault should be assigned to a corresponding role on the cluster. 

A role would consist of a name, IP and disk resource.

The qourum is backed up by the system state of the physical machines so that one can be skipped.

Userlevel 3
Badge +6

HI Sven, 

 

Based on your description, it sounds like you installed the SQL iDA on the using the cluster administrative access group (really the core cluster resource group).  That can ok assuming all of the role groups stay with which ever physical node owns that role, but if one group fails to another node you run into this situation (which is often the case).   Because of this situation, it is recommended to create a Cluster Group Client for every cluster resource group that contains data that you want to protect.

Badge +2
  • a higher log-lever doesn’t show more relevant informations.
  • a new try to backup the instance over his IP (instead of dns-name) ended with the same error.
  • backup with “use local system account” ended also with the same error
Userlevel 3
Badge +6

but thinking of that further that is maybe not the best way to solve it.

Userlevel 3
Badge +6

You, can use the Commvault process monitor (gxadmin.exe in the base folder) to higher the debug level of agents.

you can also take a look at this data interface pairs

https://documentation.commvault.com/11.22/expert/7128_dedicated_backup_network.html

Badge +2

Yes, Node A holds the cluster IP.

From Node A we can open all SQL instance running on node B (SQL Management Studio).

Is there a way to get more CV-Log-files? Debug-mode?

 

Userlevel 3
Badge +6

then its looks like that Node A is the active node of the cluster (with the cluster IP). Correct?
In that case Commvault is trying to backup also the SQL instance on Node B via Node A.

Badge +2

SQL 2017, Windows 2016

No, we don’t use always on, it’s old-school failover cluster.

Are there any special ports needed from Commvault between the two hosts? The normal SQL-Ports are open from both sites.

The dedicated user has admin-permissions on both hosts and sysadmin on every instance.

Userlevel 3
Badge +6

Hi Sven,

 

which version are you running on SQL and Windows? And are you using always on Availability groups?

be sure that there is also communication possible between the nodes and and that the backup account that you use is working with the same permissions on both nodes.

 

Reply