TrueNAS SCALE on VMware Workstation — Both Virtual Disks Suddenly Full Despite Only ~114GB of Actual Data

TrueNAS General
1

Hi everyone,

I’m running TrueNAS SCALE (Community Edition) as a VM on VMware Workstation 17.5 on a Windows host. I’m hoping someone can help me understand why my host drives are completely full, and what the best path forward is.

My Setup

  • Host OS: Windows 10
  • Hypervisor: VMware Workstation 17.5
  • TrueNAS SCALE (Community Edition)
  • Pool name: Mera
  • Pool topology: 1 x MIRROR (ZFS Mirror), 2 x 950 GiB virtual disks
    • sda → D:\ on host (931 GB physical drive)
    • sdb → E:\ on host (931 GB physical drive)

The Problem

Both host drives D: and E: are now completely full (D: has 2.67 GB free, E: has only 116 KB free), even though TrueNAS only shows ~114 GB of actual data used across all datasets.

Dataset breakdown:

  • KeToan: 63.2 GiB
  • Media: 31.4 GiB
  • LuuTru: 19.7 GiB
  • LongVT_Deploy: 9.19 GiB
  • DataBase: 944 MiB
  • Chung: 276 KiB
  • Others: small
  • Total used: ~114 GiB out of 914.91 GiB usable

What I’ve already investigated

  1. The VMDK for sda (D:) is split into 32 files (TrueNasscale-0-s001.vmdk through s032.vmdk), each ~28–29 GB → totaling ~930 GB on disk.
  2. The VMDK for sdb (E:) is a single monolithic sparse file (TrueNasscale-0.vmdk) at 928 GB on disk.
  3. I checked VMware Snapshot Manager — no snapshots exist.
  4. I tried VMware’s “Compact” utility on both disks — no change in host disk usage.
  5. I adjusted dataset quotas in TrueNAS — no effect on host disk usage (expected, since quota doesn’t shrink the VMDK).
  6. There are no other large files on D: or E: outside of the VMDK files.

Root Cause (my understanding)

It seems the VMDKs were originally provisioned as thick/split, so VMware allocated the full 950 GB upfront on the host regardless of actual ZFS usage. Now the host drives are physically full even though TrueNAS has 800+ GiB available inside the pool.

What I need help with

  1. Is there a safe way to shrink/reclaim the unused space from these VMDKs without losing data, given that I have no external storage available for a full backup?
  2. Would converting to thin-provisioned VMDKs solve this? If so, how do I do it safely with a live ZFS Mirror pool?
  3. Is there a ZFS-level operation (e.g., TRIM, zpool trim) that could help signal unused blocks back to the hypervisor?
  4. Has anyone dealt with a similar situation — ZFS Mirror on VMware with oversized VMDKs and no room to maneuver?

Constraints

  • No external drives available
  • Network shares (Y: and Z:) only have ~60–100 GB free — not enough for a full backup
  • Cannot break the mirror permanently as this is a production NAS for a small office

Any advice would be greatly appreciated. Thank you!

2

Do you have snapshots enabled in truenas? If yes take a look at how much space is used by snapshots.
But IMO you have a more serious issue, virtual disks are a big no no for virtualized truenas.
You should really use pcie passthrough of a disk controller and full disks.
See the blogpost below

1 Like
3

Please provide the output of zfs get -r used <name of pool>. I suspect it’s your DataBase dataset that’s the problem, as ZFS is copy-on-write, and doesn’t free stale blocks. Also, you need to understand the difference between ZFS snapshots and VMware snapshots.

1 Like
4

image
image513×702 16.6 KB

This is the result when the command is executed.

5

Looks like a duplicate of KeToan in a dataset named ketoan underneath is sucking up all your space.

6

image
image835×218 72.9 KB

But it only takes up 62GB; my two hard drives have a combined capacity of almost 2TB.

7

The safest way would be for you to physically replace the disks on Windows 10 machine with larger models and use software to clone over all the data. You would be able to do a single disk, 1Tb → 2Tb and verify the Windows system is using only half of the space and then perform the same for the second disk. Search for disk cloning software on the internet for different examples.

Your system should not be used for production. You need to get a solid understanding of VMWare virtualization, TrueNAS and backups. Your setup is a very good way to lose all your data.

Another issue to look at is only using CMR type hard drives and not SMR with TrueNAS / ZFS.

SMR vs CMR ServeTheHome

TrueNAS Docs - WD Red SMR Drive Compatibility with ZFS

1 Like