Event IDs 1101 and 1108 are not sufficient by themselves to diagnose the failure; capture their ProviderName, full message, timestamp, and correlated cluster/storage state while reproducing CSV creation or node maintenance.
Capture the failing operation
Run the following from elevated PowerShell immediately before reproducing the problem:
$Start = Get-Date
# Reproduce CSV creation or the node maintenance operation here.
Get-WinEvent -FilterHashtable @{StartTime=$Start} |
Where-Object Id -in 1101,1108 |
Select-Object TimeCreated, ProviderName, Id, LevelDisplayName, Message |
Format-List
Get-ClusterLog -UseLocalTime -TimeSpan 30 -Destination C:\ClusterLogs
Get-ClusterSharedVolume | Format-List *
Get-VirtualDisk | Format-Table FriendlyName,HealthStatus,OperationalStatus
Get-PhysicalDisk | Format-Table FriendlyName,HealthStatus,OperationalStatus
Get-StorageJob
The provider and complete message are essential because event numbers are provider-specific. Preserve the cluster logs from all four nodes, plus the FailoverClustering event channels. Microsoft’s storage troubleshooting guidance recommends collecting these items together so the events can be correlated with the affected CSV and virtual disk.^1^
In the cluster log, start at the event timestamp and search for:
- The CSV or cluster disk name.
-
IsAlive,Online,Offline,Move, orGroup move. -
RHS,timeout,NetFT, orSMB. - A virtual disk changing to
Detached,No Redundancy,InService, orRegenerating.
Microsoft recommends working backward from the first resource failure rather than treating later failover events as the initiating fault.^2^
Distinguish normal maintenance events from a failure
If the provider is Microsoft-Windows-FailoverClustering and the messages only report that a CSV entered and then exited maintenance, that can be expected around a maintenance operation. Suspend-ClusterResource deliberately enables maintenance for a disk or CSV; for a CSV, dependent resources are taken offline and client access is interrupted.^3^
Do not classify the events as harmless if any of the following also occurs:
- CSV remains offline or pending offline.
- The operation hangs or times out.
- A virtual disk becomes unhealthy or detached.
- Storage jobs remain suspended.
- A node is evicted or the RHS process terminates.
- CSV access or VM I/O fails.
Also distinguish CSV resource maintenance from S2D physical-disk maintenance. A physical disk in In maintenance mode has reads and writes halted; transitional states should normally progress through Starting maintenance mode or Stopping maintenance mode and return to Healthy/OK.^4^
If the problem occurs during node drain
The documented planned-drain failure pattern involves CSVs becoming unresponsive or pending offline, prolonged I/O pauses, RHS termination, and sometimes node eviction. Contributing conditions include the draining node owning quorum, file locks, NetFT packet loss, network congestion, and inconsistent adapter settings that prevent SMB Multichannel connectivity.^5^
If your captured logs match that pattern:
- Move quorum ownership and critical cluster roles away from the node before draining it.
- Compare network-adapter configuration across all four nodes.
- Check the cluster log for NetFT packet loss and SMB Multichannel failures.
- Check for file locks against the affected CSV.
- Run cluster validation and review the Storage Spaces Direct and network sections.
Since controller drivers, firmware, and Windows updates are already current, those checks do not need to be repeated. Cluster validation remains useful because it tests the active configuration and verifies that cache devices and S2D storage are reported without errors.^6^
Do not apply recovery overrides unless the state matches
Commands involving DiskRecoveryAction, DiskRunChkDsk, removing a disk from CSV, or forcing it online are recovery procedures—not general troubleshooting steps. Use them only if Get-VirtualDisk specifically reports No Redundancy or Detached, following the matching Microsoft procedure.^6^
If every reproduction causes an actual CSV or node-maintenance failure while the virtual disks remain Healthy/OK, submit the cluster logs, exported FailoverClustering logs, validation report, and exact 1101/1108 provider messages to Microsoft Support. Those details are required to distinguish an S2D resource failure from a network, SMB, filter-driver, or cluster-orchestration issue.
References
- Troubleshoot storage issues in Hyper-V and Windows Server failover clusters
- Unexpected cluster failover troubleshooting guidance
- Suspend-ClusterResource (FailoverClusters) | Microsoft Learn
- Troubleshoot Storage Spaces and Storage Spaces Direct health and operational states
- Shared Volumes Don't Respond During Planned Cluster Node Drain - Windows Server | Microsoft Learn
- Troubleshoot Storage Spaces Direct