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

NetLeverage
NetLeverage

VM stops with "There is no more space for virtual disk": free the datastore before you answer Retry or Cancel

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.

When a datastore runs out of room, the VM stops responding and the vSphere Client shows a question that begins "There is no more space for virtual disk". The prompt continues "You might be able to continue this session by freeing disk space on the relevant volume, and clicking Retry. Click Cancel to terminate this session." Do not answer it yet. First find out what kind of datastore is full, free or add space, and only then click Retry.

What the symptom means

Broadcom KB 318352 (Updated On: 11-12-2026, ESXi 7.x and 8.x) says that a virtual machine "stops responding when a datastore exhausts its physical capacity", and that this "often occurs during thin-provisioned disk growth or snapshot consolidation". The KB lists these markers: the VM's vmware.log contains msg.hbacommon.outofspace and the line "There is no more space for virtual disk", the host's /var/run/log/vmkernel.log contains "No space left on device", and the UI shows an error message. KB 409314 places the prompt under the VM's Issues and Alarms tab, and KB 326433 shows it as a warning on the VM summary.

The KB's stated cause on VMFS is that the datastore has used all its small and large file blocks, so new block allocations for VM writes and swap files stop. Snapshot consolidation also needs temporary working space to merge delta disks into the base disk, so it fails when that space is missing.

The same prompt appears on other storage, with different causes:

  • NFSv3 datastore: KB 409314 (Updated On: 07-17-2026) says the vSphere Client may show free space while the storage array's physical pool is exhausted. The fix is on the array.
  • vSAN: KB 326433 (Updated On: 04-17-2026) found individual capacity disks full while cluster-wide space looked adequate, and KB 445473 (Updated On: 06-26-2026) found many snapshot disks filling the vSAN datastore.

Likely causes

Cause What you see How to confirm
VMFS datastore full from thin-disk growth, snapshots or swap files Prompt; msg.hbacommon.outofspace in vmware.log; "No space left on device" in vmkernel.log esxcli storage filesystem list lists the volume; the Host Client Storage view (Capacity, Free Space, Used Space) shows free space near zero
Snapshot consolidation without working space The prompt appears during or after consolidation Snapshot chain size against free space; "Snapshot consolidation fails: not enough space on the datastore" covers this case
NFSv3 datastore: backend array full though the datastore shows space A warning ending "No space on NFS volume." in vmkernel.log Physical pool or aggregate utilisation in the storage management interface (KB 409314)
vSAN capacity disks full or many snapshot disks Prompt on many VMs; vSAN Skyline Health reports a 'Disk Capacity' error esxcli vsan debug disk overview list and esxcli vsan debug object list (KB 445473)
Swap file too large Power-on fails on space VM memory size against free space; KB 318352 suggests a temporary memory reservation

Prove the cause before changing anything

All of these only read.

1. Read the question. In the vSphere Client the prompt is on the VM's summary or under Issues and Alarms. The VM name and path in it name the datastore volume that is full.

2. Check the datastore free space on the host.

esxcli storage filesystem list

KB 381759 (Updated On: 02-24-2026, ESXi 7.x and 8.x) uses this command to list mounted volumes. The KB's output example is not available as text, so read the free-space figure for the volume named in the prompt in the Host Client (KB 405611 below), and in the command's output if it shows one. A good result has room; a bad result has free space near zero. For the browser view, KB 405611 describes the Host Client route: Storage, select the datastore, read Capacity, Free Space and Used Space.

3. Read the logs for the VM and host. The VM's vmware.log is in its folder under /vmfs/volumes/ and the host's /var/run/log/vmkernel.log. Look for the three markers above.

4. Read the pending question from the host command line, if you need it. KB 311492 documents vim-cmd vmsvc/getallvms to list virtual machine IDs and vim-cmd vmsvc/message followed by the VM ID to display the question and its choices. That KB is written for ESXi 5.5 to 6.7 and for a different question (moved or copied), so check the output on your host before using it for answering. Answering is a change; click Retry in the vSphere Client unless you cannot reach it.

5. For vSAN, run the two esxcli vsan debug commands above. KB 445473 says they identify the snapshot disks and their usage.

Fix

The aim is to give the datastore real room, then answer Retry.

Do not click Cancel first. The prompt says Cancel terminates the session. KB 318352 and KB 409314 do not say what state the guest is left in afterwards, so free space first.

Free space on VMFS

Risk: a Storage vMotion copies the VM's disks to the target datastore and puts load on both datastores while it runs. Save first: record the VM's current datastore and the target's free space. Back out: a completed move can be reversed with another Storage vMotion back to the original datastore, if that datastore still has room.

Risk: a temporary memory reservation takes memory from the host for that VM, which can affect other VMs; KB 318352 suggests it when the swap file is too large. Save first: note the VM's current reservation value. Back out: set the reservation back to the noted value when the space problem is solved.

Risk: deleting files cannot be undone. Save first: list the files in the datastore with their sizes (Broadcom's KBs do not describe a tool for this, so use the datastore file view in your vSphere Client or Host Client), write down the exact file you will remove, and confirm that no VM or template uses it. Remove one named file at a time. Back out: none for a deleted file, which is why the check comes first.

Risk: expanding a datastore changes storage that other VMs also use. Save first: record the datastore's current capacity and free space from the Host Client Storage view. Follow Broadcom's "Increase Capacity of vSphere VMFS Datastore" page, which KB 318352 links; "Grew the LUN but the VMFS datastore did not get bigger: rescan, then Increase Capacity" also covers the steps. Back out: no step here shrinks a datastore again, so agree the new LUN size with your storage team before you start.

Risk: snapshot consolidation and deletion run for a long time and need working space; KB 318352 says it creates a temporary snapshot when the VM is powered on and needs no extra space when the VM is powered off, but the same KB adds that thin-provisioned base disks expand as snapshot data is committed, up to the maximum Provisioned Size, so power-off does not remove the need for free space. Wait for any running consolidation to finish before starting another. Save first: note the VM's snapshot names and sizes from the Snapshot Manager. Back out: committing a snapshot into the base disk is not reversed by any step here, so start it only after the datastore has the buffer below.

KB 318352 lists these ways to get space, with notes on each; each has its warning above:

  • Clear unwanted data such as old log bundles and ISO images.
  • Expand the LUN on the storage array and increase the VMFS datastore size, then rescan storage. KB 318352 recommends a cluster-level rescan on all clusters the datastore is connected to.
  • Consolidate or delete existing snapshots. A 20% free space buffer is recommended for stable thin provisioning and consolidation.
  • Storage vMotion the VM to a datastore with more space.
  • Add a temporary memory reservation if the swap file is too large.

KB 318352 also records that Storage vMotion can fail with the same pending question, so a full datastore may block this route.

NFSv3 backend full

KB 409314 says to check actual utilisation in the storage provider's interface, delete unused snapshots, old volumes or orphaned files on the array, or increase the physical pool. These actions happen on the storage array, and they belong to the storage owner.

Risk: deleting a snapshot, volume or file on the array cannot be undone, and an array volume or export may be used by other hosts or by replication. Save first: record the name, owner, size and last use of each object, and ask the storage owner to take an array snapshot or export a list of the objects before anything is removed. Select one named object at a time and confirm that no VM, template or replica on any host uses it. Back out: none for a deleted object, other than the storage owner's array snapshot or backup. When the storage owner confirms physical space, click Retry in the vSphere Client prompt or retry the task.

vSAN

KB 445473 says to add capacity (disks or hosts), clear unnecessary data, and let the resync move data from high-usage disks. Consolidate the VM only after the datastore "has at least 1.5x the size of the total snapshot data". KB 326433 advises adding capacity or migrating VMs to another cluster or non-vSAN datastore. It also says to identify and delete unused or non-critical VMs or VMDKs, and to clean up unassociated vSAN objects only "after verifying the prerequisites outlined in the document Procedures for identifying Unassociated vSAN objects".

Risk: deleting a VM, a VMDK or a vSAN object cannot be undone, adding capacity changes the cluster, and a migration puts load on the cluster while it runs. Save first: record the name, owner, size and last use of each VM or object, confirm with its owner that it is unused, and have a current backup of anything you are not certain about. Select one named object at a time, after reading the KB's identification procedure for unassociated objects. Back out: none for a deleted object other than a restore from backup; a completed migration can be reversed with another migration if the original location still has room.

Answer the question

When free space exists, click Retry in the vSphere Client. KB 409314 says to click Retry in the prompt once physical space is available.

Example (illustrative): app01 runs on ds-prod01, a 2 TiB VMFS datastore (2 × 1,024 = 2,048 GiB). Thin disks and a forgotten snapshot grow until app01 stops with the prompt. The Host Client shows free space for ds-prod01 near zero, and esxcli storage filesystem list lists the volume. The administrator finds a 40 GiB old log bundle and two ISO images that nobody uses, confirms them, and removes them, but that is not enough for the 20% buffer: 20% of 2,048 GiB is 409.6 GiB. The storage team grows the LUN by 512 GiB to 2,560 GiB, so the buffer needed is 20% of 2,560 = 512 GiB. After the VMFS datastore is increased and storage rescanned, the free space is above that figure and the administrator clicks Retry.

Confirm the fix

  1. The Host Client (and esxcli storage filesystem list, if its output shows free space) shows free space for the volume at or above your chosen buffer. KB 318352 recommends 20% free.
  2. The VM answers on the network and its console, and the question is gone from Issues and Alarms.
  3. No new msg.hbacommon.outofspace or "No space left on device" lines appear in the logs.
  4. Plan the consolidation or snapshot clean-up that caused the growth ("The -000001.vmdk and delta files in a VM folder: what each one is" explains the files involved).

What to watch afterwards

  • Alert on datastore free space before the 20% mark, and on thin-provisioned over-commitment.
  • Watch snapshot age and size; KB 445473 ties vSAN exhaustion to a large number of snapshot disks.
  • For NFS, alert on the array pool, not only the datastore figure.

Risks and limits

  • These commands have not been run or lab-tested.
  • KB 311492's vim-cmd vmsvc/message page covers ESXi 5.5 to 6.7. Broadcom's KB does not show that command's output on ESXi 7 or 8, so no answer command is given for this prompt.
  • Broadcom's wording is that the VM "stops responding"; the vSphere Client may show it waiting for an answer to the question above.
  • The 20% buffer and the 1.5 times figure are the KBs' recommendations for their cases, not a universal rule.

Sources