Hello
We are running a two nodes fail-over cluster on Windows 2012 R2 running four SQL Server instances and connected to network via a converged network adapter HP CN1200E for LAN communications and iScsi LUN access.
A couple of days ago the cluster service failed on node 2 while all instances were running on it.
Event logs show that issues occurred with some Windows cluster resources:
Event 1215 was raised on node 2.
Cluster network name resource 'Cluster Name' failed a health check. Network name 'PROD-SQL-BU' is no longer registered on this node. The error code was '-1073741663'. Check for hardware or software errors related to the network adapter. Also, you
can run the Validate a Configuration wizard to check your network configuration.
All cluster resources then failed and all SQL Server instances were activated on node 1.
I have to admit I am a bit puzzled out.
I am NOT a beginner regarding fail-over clusters however I have to admit I do not know what to do to solve the issue with that cluster.
This cluster runs many resources as you can see in the summary below:
> Role "SQL Server (AX)"
- Role:Analysis Services (AX)
- Storage:3 LUN
- Server name :PSBU(IP address x.x.x.36)
- Other ress. :SQL Server (AX), SQL Server Agent (AX)
> Role "SQL Server (DWH)"
- Role:Analysis Services (DWH)
- Storage:3 LUN
- Server name :PSBU-DWH(IP address x.x.x.196)
- Other ress. :SQL Server (DWH), SQL Server Agent (DWH)
- File server :\\PSBU-DWH
> Role "SQL Server (M1)"
- File server :\\PSBU-M1
- Storage:3 LUN
- Server name :PSBU-M1(IP address x.x.x.37)
- Other ress. :SQL Server (MISC1), SQL Server Agent (MISC1)
> Role "SQL Server (SP)"
- Storage:3 LUN
- Server name :PSBU-SP(IP address x.x.x.127)
- Other ress. :SQL Server (SP), SQL Server Agent (SP)
As I said earlier network access is made using a Converged network Adapter HP CN1200E.
The card has an "iScsi" personnality allowing both classic LAN communications and iScsi communications.
Windows Server sees both 10 GBits ports and a Team is created using both 10GBits CN1200E ports in switch independent mode.
Another 1Gbits adapter connection on each node is made to allow cluster communications between nodes using a non-routed IP address.
(This is old fashioned but was set up to prevent complete loss of communications between nodes if the CN1200E card fails).
All monitoring software and probes showed that the network interfaces of CN1200E adapter did not failed.
Cisco logs also showed that network communication was never lost with the 10 Gbits adapter.
CLUSTER VALIDATION REPORT
I ran the report this morning.
It showed the configuration was not correct with a specific settings of each cluster role : for each role the report advised that I should set parameter "RegisterAllProviderIP" to 0 to prevent issues.
I changed that setting and it will be applied next time the roles move from one node to another.
QUESTIONS
What should I also check?
Florent