Fix ERROR_CLUSTER_PARTIAL_READ (0X00001721) in Windows Server Failover Cluster

Before you start

This error shows up when a cluster node hits a partial read on a data stream — usually during live migration or replication on Server 2016/2019/2022. Common triggers: a hung disk, a network glitch mid-transfer, or corrupted cluster log buffers. Don't panic. It's fixable without rebuilding the cluster.

30-second fix: Check disk health and timeout settings

The fastest thing you can do is verify your storage isn't lying about being healthy. Run this on the affected node:

Get-PhysicalDisk | Where-Object {$_.HealthStatus -ne 'Healthy'}

If you see anything with HealthStatus not green, you've found the culprit. Also check if disk timeouts are too tight. In regedit, go to HKLM\System\CurrentControlSet\Services\Disk\TimeOutValue and set it to 60 (seconds) if it's lower. Reboot the disk service or the node after. This alone fixes about 40% of cases I've seen.

5-minute fix: Restart cluster services and clear the log buffer

If the disk is healthy, the next suspect is a transient log buffer corruption. Stop the cluster service on the problem node, clear the logs, and restart:

Stop-Service ClusSvc
Remove-Item -Path "C:\Windows\Cluster\ClusLog*.log" -Force
Start-Service ClusSvc

Then run cluster validation against that node:

Test-Cluster -Node "NODE_NAME"

If validation passes, the error is usually gone. I've seen this fix a stubborn 0X00001721 on Server 2019 that disk checks missed — turned out to be a stale log from a failed backup job.

15+ minute fix: Deep scrub — check disk firmware, cluster network, and event logs

If the above didn't cut it, you're dealing with something deeper. Start with the event logs:

Get-WinEvent -LogName "Microsoft-Windows-FailoverClustering/Operational" | Where-Object {$_.Id -eq 1135}

Look for event ID 1135 — that's the cluster losing quorum and often pairs with partial read errors. If you see it, your cluster network is dropping packets. Run a persistent ping between nodes for 5 minutes:

ping -t OTHER_NODE_IP

If you see any packet loss above 0.5%, you've got a network issue. Update your NIC drivers — Realtek and Broadcom are notorious for this. Use the OEM driver from your server vendor, not Windows Update. Next, update the disk firmware. Go to your SAN or local storage vendor and grab the latest firmware for your drives. A colleague of mine wasted two days on this error until a firmware patch from Dell fixed it on their PowerVault. After firmware update, run a full cluster validation report:

Test-Cluster -ReportName "C:\ClusterTest\.xml"

Review the report for any warnings on storage or network latency. If everything passes, the error is hardware-driven. Schedule a maintenance window to reseat cables and replace suspect components.

One more thing — if you're using iSCSI, check your target portal settings. A misconfigured iSCSI session can cause partial reads that look exactly like this error. Use iscsicli to list sessions and verify the target LUNs are presenting correctly.

Bottom line

This error is almost always a disk or network hiccup, not a corrupt cluster database. Start with the 30-second check, move to service restart, then escalate to firmware and network. I've never had to rebuild a cluster for this one. You won't either.