Fog Project Documentation (FOG Docs)

Deploy Refused — Sector Size Mismatch

Properties3
descriptionWhat to do when a deploy stops with a "Sector size mismatch" message
tagstroubleshooting, imaging, deploy, nvme, 4kn
aliasesSector Size Mismatch, Deploy Refused Sector Size

3 min read

Permalink

Edit this page · View source

Deploy Refused — Sector Size Mismatch

Symptom

A deploy task stops early — before the disk is written — with a message like:

Sector size mismatch
   Image was captured on a disk with 4096-byte logical sectors, but /dev/sda
   uses 512-byte logical sectors.
   Partition-table and filesystem geometry cannot be translated between logical
   sector sizes, so this image cannot be deployed to this disk.
   Deploy this image only to a disk with 4096-byte logical sectors, or capture a
   new image on a disk with 512-byte logical sectors.

What it means

The image was captured on a disk whose logical sector size (512 or 4096 bytes) is different from the disk you’re deploying to. FOG cannot translate partition-table and filesystem geometry between the two, so it refuses rather than write a disk that won’t boot. Nothing was written — the target disk is untouched.

On newer FOS builds the message also includes a line about the target’s device type, for example:

   /dev/mmcblk0 is an eMMC/SD device; its 512-byte logical sector size is fixed
   by the MMC/SD specification and cannot be changed. Only an image captured on
   512-byte-sector hardware can deploy to it.

That line tells you which side of the mismatch can be fixed: if the target’s sector size is fixed (eMMC/SD, UFS), recapturing on matching hardware is the only remedy; if the target is a virtual disk, you can change its sector size in the VM’s disk configuration instead.

For the full explanation of why this happens — and what each device type can and can’t do — see Sector Sizes and Imaging.

How to fix it

Pick whichever fits your situation:

  1. Deploy to matching hardware. Send this image to a disk with the same logical sector size it was captured on (the message tells you which). 512n and 512e disks are interchangeable; 4Kn disks are not interchangeable with either.

  2. Recapture on the target geometry. If you need this image on the current hardware, capture a fresh image from a machine whose disk matches, then deploy that image instead.

  3. Let FOS reformat an NVMe target. If the target is an NVMe drive that supports the image’s sector size, FOS offers to low-level reformat it to match after a 60-second cancel window. See NVMe targets can be reformatted to match. NVMe is the only device type this works for — not SATA/SAS, eMMC/SD, UFS, or USB targets, and not NVMe drives that are 4Kn-only. See Sector sizes by device type for the per-type breakdown.

  4. Virtual machine target? The disk’s logical sector size is set by the hypervisor, not the virtual disk itself. Change the disk’s logical_block_size (QEMU/libvirt/Proxmox) to match the image and redeploy.

Checking a disk’s sector size

From a FOS shell (a debug task) or any Linux host:

blockdev --getss /dev/sdX     # logical sector size: 512 or 4096
blockdev --getpbsz /dev/sdX   # physical sector size

Match the logical size (--getss) between the image’s source disk and the deploy target.

See also

Graph View

Backlinks