Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

Storage vMotion to vSAN fails for a VM with snapshots and disks on different storage policies

Technical article
VMware

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

Broadcom lists a known issue in vSphere 8.0 Update 3: moving a virtual machine to a vSAN datastore with Storage vMotion "might fail" when the VM has at least one snapshot and more than one virtual disk with different storage policies. The release-note workaround is to split the move: first the VM home and one disk, then the remaining disks in a disk-only migration. A second Broadcom article reports that this second step can itself fail and orphan the VM, so read the Fix section before starting. Before using it, prove the VM matches all three conditions and rule out the other documented reasons a move to vSAN fails, because several of them look the same from the task list.

What the error means

The release note heading is "vSphere Storage vMotion operations might fail in a vSAN environment due to an unauthenticated session of the Network File Copy (NFC) manager". The text explains the trigger and the cause (VMware vCenter 8.0 Update 3k Release Notes, under "Known Issues from Previous Releases"; the wording is the same in 3i and 3j):

"Migrations to a vSAN datastore by using vSphere Storage vMotion of virtual machines that have at least one snapshot and more than one virtual disk with different storage policy might fail. The issue occurs due to an unauthenticated session of the NFC manager because the Simple Object Access Protocol (SOAP) body exceeds the allowed size."

In plain terms: the copy service that moves the disk files sends a request describing the VM's disks, snapshots and policies. With a snapshot and several policies, that request grows past an allowed size, the copy session is not authenticated, and the move fails. The same paragraph appears under "Known Issues from Previous Releases" > Storage Issues in the VMware ESXi 8.0 Update 3k Release Notes.

The release note does not print the error message the vSphere Client shows. That is why this fault is matched by its three conditions, not by text. The conditions are:

  1. The destination is a vSAN datastore.
  2. The VM has at least one snapshot.
  3. The VM has more than one virtual disk, and the disks do not all use the same storage policy.

Snapshots on their own do not block Storage vMotion. The vSphere 8.0 storage guide says: "You can migrate VMs with snapshots across different datastores" (VMFS Datastore Concepts and Operations).

Likely causes

A failed move to vSAN for a VM with snapshots has five documented causes. Each needs a different fix. Rows 3 to 5 are Storage vMotion failures to any destination, not only vSAN; rule them out first.

Cause What you see How to confirm
The NFC known issue (snapshot + several policies + vSAN target) Move to vSAN fails; the VM meets all three conditions Snapshot list, disk list and the policy on each disk (below)
Part of the VM already on the target vSAN "A general system error occurred: Timed out waiting to capture VM ID at destination", at 100% Some disks already on that vSAN datastore; VM may appear under "Discovered VMs" afterwards (KB 458676)
Missing change-tracking file "Storage vMotion failed to initialize copy of disk-specific file." ctkEnabled = "TRUE" in the .vmx and a "Could not find the file" line for -ctk.vmdk in vmware.log (KB 395402)
A stale backup lock "The method is disabled by '…'"; migrate option greyed out The name in quotes is a backup product (KB 315486)
Not enough space for the snapshots Failure late in the copy, often at 90% or more Free space against 1.5 times the snapshot sizes (KB 318799, KB 318856)

On a vSAN ESA source cluster, also read the separate KB 377763 before moving any VM with snapshots off it (see Risks and limits).

Prove the cause before changing anything

1. List the VM's snapshots. On the host running the VM, KB 316575 (Consolidating/Committing snapshots in VMware ESXi) gives:

vim-cmd vmsvc/getallvms
vim-cmd vmsvc/snapshot.get <Vmid>

Any snapshot in the output meets condition 2. Snapshot Manager in the vSphere Client shows the same thing.

2. List the disks and where they live. Also from KB 316575:

grep vmdk /vmfs/volumes/<datastore>/<vm_folder>/<vm_name>.vmx

Each scsiX:Y.fileName line is one disk. A name ending -000001.vmdk is a snapshot child disk. Note any disk path that is already on the target vSAN datastore.

3. Check the storage policy of every disk. In the vSphere Client, open the VM's Edit Settings: each hard disk shows its VM storage policy. If there is more than one disk and the policies differ, condition 3 is met.

4. Check for a part of the VM already on the target. KB 458676 (Storage vMotion to vSAN Fails with "Timed out waiting to capture VM ID at destination") describes the case where "the virtual machine currently has virtual disks distributed across mixed storage (e.g., some on VMFS and others on vSAN)" and the remaining disks are being moved to the same vSAN datastore. Its cause: "the migration destination (vSAN) already contains a directory for that specific virtual machine". If step 2 shows disks already on the target vSAN, this is the more likely match. The split-move workaround below creates that state itself.

5. Check change tracking. KB 395402 (Storage vMotion failed to initialize copy of disk-specific file) shows the .vmx lines ctkEnabled = "TRUE" and scsi0:0.ctkEnabled = "TRUE", and vmware.log lines such as "Disk_SVMotionPrepareCBT: Failed to get CTK file maximum size … Could not find the file(393218)". Search for them:

grep -i ctk /vmfs/volumes/<datastore>/<vm_folder>/<vm_name>.vmx
grep -i "SVMotionPrepareCBT" /vmfs/volumes/<datastore>/<vm_folder>/vmware.log

6. Check for a backup lock. KB 315486 (The method is disabled by 'SYMC-INCR dd-mm-yyyy hh:mm') lists messages such as "The method is disabled by 'Zerto Source'", with "Storage vMotion option is grayed out or unavailable." Its cause: backup software disables migration during a backup, and the entries are "not properly cleared after the backup task completes or fails".

7. Check space. KB 318799 (Storage vMotion tasks are failing for powered on VMs with multi-writer disks) says: "Snapshots of the virtual machine will impact storage vMotions if space on the LUN is less than 1.5 times the total of the VM's snapshots sizes." KB 318856 adds that "Storage vMotion or migrations typically fail at 90% or higher when the destination datastore lacks sufficient space for the virtual machine or its overhead files." Compare the free space on the destination with the size of the VM plus 1.5 times its snapshot files.

If the VM meets all three conditions in steps 1–3, and steps 4–7 are clean, the known issue is the documented match.

Example (illustrative): app07 has three disks on a VMFS datastore. Disk 1 and the VM home use "Gold", disks 2 and 3 use "Silver". It has one snapshot from last night's backup. Moving it to vsan-ds01 fails. No part of app07 is on vsan-ds01, there is no CTK warning and no "method is disabled" message, and there is room. All three conditions are met, so the split move is the documented route, with the KB 458676 recovery ready.

Fix

The known issue: split the move. The release notes' workaround: "First migrate the VM home namespace and just one of the virtual disks. After the operation completes, perform a disk only migration of the remaining 2 disks." With more than three disks, move all the rest in the second step.

  • Step one: right-click the VM > Migrate > Change storage only > Configure Per Disk; tick the VM home and one disk, choose Configure, pick the vSAN datastore, choose Confirm, then finish the wizard (KB 430912). Leave the other disks where they are. Risk: this leaves part of the VM on the target vSAN datastore, which is the state KB 458676 describes below.
  • Step two: once that completes, migrate the remaining disks to the vSAN datastore. Risk: KB 458676 reports that this move "may fail at 100%" with "Timed out waiting to capture VM ID at destination", and that the VM "may become orphaned and appear under 'Discovered VMs'" while the guest keeps running. The vCenter and ESXi release-note workaround and KB 458676 therefore conflict: the workaround's second step is the case KB 458676 documents as failing. If step two fails, use the KB 458676 recovery in the next section.

Risk: between the two steps the VM's storage is split across two datastores. Keep the window short and do not take new snapshots in between.

A second route follows from the release note's own conditions, though it is not its stated workaround: remove one of the conditions. Deleting the snapshot (after the backup that created it has finished with it), or setting all disks to the same storage policy, means the VM no longer matches. Risk: deleting a snapshot consolidates its changes into the base disk and cannot be undone. Agree with the backup owner first.

Part of the VM already on vSAN. KB 458676: "Migrate the virtual disks off the target vSAN datastore to a temporary non-vSAN (VMFS) datastore so that no part of the VM remains on the target vSAN." Then "perform a single Storage vMotion to move the entire virtual machine to the target vSAN datastore in one operation." Disruptive step: this moves the disks twice; plan for the extra copy time and space. If the VM has become orphaned under "Discovered VMs", deal with that first: KB 458676 notes the guest "remains running and reachable".

Missing change-tracking file. KB 395402: "Disable CBT on the virtual machine", "Re-initiate the Storage vMotion", then enable CBT if needed "and confirm that the ckt.vmdk file is created under the home folder." Risk: backup products that rely on CBT need a new full backup after CBT is reset.

Stale backup lock. KB 315486 Option 1: "Rerun or schedule a new backup task for the affected virtual machine." A successful backup normally clears the entry. KB 315486's Option 2 edits the vCenter database. Risk: the KB says to "Take offline snapshots of all vCenter Server Appliance (VCSA) instances" first.

Not enough space. Free space on the destination, or consolidate the snapshots first. KB 318799's fallback where the LUN cannot grow is to clone the disk to another datastore with vmkfstools -i. Disruptive step: the VM must not be writing to the disk being cloned, and the clone must be attached in place of the original.

Confirm the fix

  • The Migrate task completes and the VM's Datastores tab shows only the vSAN datastore.
  • grep vmdk on the new .vmx shows every disk on the vSAN path.
  • Every disk shows the intended storage policy, and it shows compliant.
  • If CBT was reset, a -ctk.vmdk exists beside each disk after the next backup.
  • No "Discovered VMs" entry remains.

What to watch afterwards

  • Snapshots before moves. Check for snapshots before any planned Storage vMotion to vSAN; backup snapshots that are left behind meet condition 2.
  • Release notes. Check later vCenter and ESXi 8.0 Update 3 release notes for this known issue moving to the resolved list.
  • Backup locks after failed backup jobs, which leave "method is disabled" entries.

Risks and limits

  • The release notes describe the fault as one that "might fail"; the absence of a named error message means the match is by conditions. Collect the vCenter and host logs for the failed task if the split move also fails, and open a support case with those logs.
  • The release notes do not name a fixed build for this issue as of 17 November 2026; it is still listed in vCenter 8.0 Update 3i, 3j and 3k and ESXi 8.0 Update 3k.
  • Conflicting vendor guidance. The release-note workaround's second step is the case KB 458676 documents as failing. Treat both steps as risky and have the KB 458676 recovery ready.
  • vSAN ESA sources. KB 377763 (Migration of VM's with snapshots from a vSAN ESA 8.x Datastore may result in inconsistencies) says that on vSAN ESA before 8.0 U3b, moving VMs with native snapshots off an ESA datastore "may result in inconsistencies", including "the VM failing to boot up at the destination". Broadcom "recommends upgrading vSAN ESA clusters to VMware-ESXi-8.0U3h-25067014 (80P07) and above". Check the source cluster's build first.
  • KB 318799's main subject is multi-writer disks; its 1.5 times rule is from that page's additional information.

NetLeverage runs its own hosting on VMware vSphere, vCenter and ESXi. Talk to an engineer.

Sources