Failover Cluster
Overview
A Failover Cluster Instance (FCI) moves between the nodes of a Windows Server Failover Cluster. When something goes wrong the first questions are always the same: which node is it on now, where could it go, when did it last move, and why?
The Failover Cluster report answers them on one page:
- Nodes – every node that can own the instance, drawn as a box, with the current owner highlighted. A node that is down or paused is flagged, and so is an instance with nowhere left to fail over to.
- Quorum – the Windows cluster name, the quorum model and state, each member’s vote and the witness. Forced quorum is a problem; an even number of voting nodes with no witness is a warning.
- Properties –
FailureConditionLevel,HealthCheckTimeoutandVerboseLoggingfromsys.dm_os_cluster_properties, each compared with the default (3, 60,000 ms and 0) and explained. - Shared drives – the drives and Cluster Shared Volumes the instance depends on.
- Health – the latest
sp_server_diagnosticsstate for system, resource, query_processing, io_subsystem and events, plus the warning and error results thesystem_healthsession still holds. - Failover history – every start the error logs still record, on a timeline. The instance writes the node it started on near the top of each new log, so two starts in a row on different nodes are a failover.
On an instance with Always On availability groups but no FCI, the report shows the quorum, the cluster’s nodes (with the node this instance runs on highlighted) and the health panel. On an instance that is neither, it says Not a clustered instance and reads nothing else.
Where to find it
| Route | How |
|---|---|
| Server tree | Right-click the server → Instance Level Reports → Failover Cluster |
| Instance reports navigator | Recovery group, after Availability Groups |
| Related Links bar | From Availability Groups, Error Log and Host and Hardware |
| Report arrows | Previous is Failed Jobs, next is Failover Compatibility |
Reading the page
The diagram at the top has four parts. Click any node, the quorum box, a health tile or a point on the timeline to select its row in the grid; hover it to read the row in full.
| Status | Meaning |
|---|---|
| OK (green) | Healthy, or the default setting |
| Warning (amber) | Worth a look: a paused node, a non-default setting, a past health warning, a failover |
| Problem (red) | A node down, forced quorum, FailureConditionLevel 0, a component in error |
| Info (gray) | Information only |
| Unavailable | This login or this version could not read it; the notes say why |
Double click a row that names a report (quorum rows open Availability Groups, history rows open the Error Log) to go there. The grid supports the usual CSV and Excel export, and Copy as Markdown copies every row for a write up.
Permissions and versions
- The node, property, drive, quorum and health views need VIEW SERVER STATE (VIEW SERVER PERFORMANCE STATE on SQL Server 2022 and later). Without it those rows are marked unavailable.
- The failover history reads the error logs with
xp_readerrorlog, the same way the Error Log report does. A login that cannot read the error log loses the history, not the page. sys.dm_os_cluster_properties,sys.dm_hadr_clusterandsp_server_diagnosticsarrived in SQL Server 2012; the node status columns andsys.dm_io_cluster_valid_path_namesin 2014. On older builds those parts are marked unavailable.- SQL Server only reports the quorum when availability groups are enabled on the instance. For an FCI without them, Failover Cluster Manager shows it.