Discovery and SQL Clusters
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎12-06-2018 05:31 AM
Hi,
We notice that Discovery when discovering an MSFT SQL Instance CI on a SQL cluster names the instance including the hostname of the cluster node the instance is running on at the time of discovery.
For instance if:
Instance name: I01
Cluster name: sqlcluster1
Node A: sqlserver1
Node B: sqlserver2
And I01 is currently running on sqlserver1 then Discovery will name the instance I01@sqlserver1 and create RUNS ON relations to sqlserver1 as well as sqlcluster1.
What happens if I01 fails over to sqlserver2 on the next Discovery scan. Will Discovery then rename the instance to I01@sqlserver2 and create a new RUNS ON relation to that server, or will Discovery keep the original name I01@sqlserver1?
How can we instead configure Discovery to name the instance I01@sqlcluster1 thus matching the name of the cluster?
Best Regards,
Kristian
- Labels:
-
Discovery
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎12-06-2018 08:44 AM
That is correct, a duplicate is created (but not true duplicate because server name is different.)
I've had an enhancement request in for over two years to fix this. It really should be using the Cluster Role rather than the Cluster or the Node. Since the Cluster Role is where the actual configurations happen.
Our consultants tried several fixes but the side effects were not acceptable because they all affected other application discovery not in Clusters.
We needed up creating a Business Rule that only allowed ecmdb admin or admin role to see CIs with the Asset Tag populated with "Invalid" then CMDB was monitored closely for those duplicates and the asset tag populated with Invalid. We also populated the Description field with specific text to indicate the appropriate CIs for usage in INC, CHG and Requests.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎01-03-2019 11:35 PM
I suppose a way to circumvent this could be manually to create RUNS ON relations from the instance down to both server CIs running in the cluster.
Then Discover will not create the duplicate instance CI in case the instance fails over to the other server node.
And to avoid Discovery naming the instance after the server a reconciliation rule can be created on the 'Name' field ensuring the manually updated name reflecting the name of the cluster instead of the server is not overwritten by Discovery.
Anybody having a better solution to this?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎07-27-2020 01:34 PM
If you are still active on the Community, could you provide me a sample of what your BR looked like and or whether you are still contending with the problem described.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
‎07-27-2020 12:57 PM
Can anyone on this thread tell me if a solution was ever achieved?