Skip to main content

btrfs and ext4 data disks

If some or all of your Unraid data disks use btrfs or ext4 instead of XFS, they migrate the same way: Hoserva adopts each disk as it is, without formatting it. This page explains how a disk passes, which checks can refuse one, and what to do when one does.

Unraid is a trademark of Lime Technology, Inc.; Hoserva is not affiliated with it.

What does adopting a btrfs or ext4 disk mean?​

Hoserva supports XFS, ext4 and single-device btrfs data disks. Each disk is checked and adopted on its own, so one array can mix all three.

  • No formatting. The disk keeps its filesystem and every file on it. Your data disks are never formatted by the migration.
  • Read-only until the point of no return. The scan and the import mount each disk read-only. The verify then compares every adopted disk with what the scan recorded.
  • The same steps. Nothing else in The migration changes for these disks.

A disk that fails a check is refused by name and the scan carries on with the others. A refused data disk makes the verdict No-go, and the Review step marks its row Not adopted.

What does the scan check on each disk?​

The scan runs the checks below before it mounts anything. They read the disk and never write to it.

  1. The filesystem is what Unraid says it is. The Flash Backup names a filesystem for each disk, and the disk itself reports one. They have to agree.
  2. The filesystem is a single device. A btrfs filesystem that spans several disks is refused. See Unsupported arrays.
  3. The filesystem has an ID of its own. Hoserva identifies a filesystem by its UUID, a unique ID, so two data disks cannot share one.
  4. The log is clean. A btrfs or ext4 filesystem keeps a log of recent changes. If the last stop left changes in it that were never applied, the disk is refused.
  5. The filesystem passes a read-only check. The scan runs e2fsck -n on an ext4 disk and btrfs check --readonly on a btrfs disk, on the unmounted device. Neither command changes anything on the disk.

The scan reports these under Data disk filesystem and Filesystem check. A disk that passes both is listed as a single btrfs or ext4 filesystem with a clean check.

What do the refusals mean?​

The report states each refusal in words. The table gives its meaning and the fix. If you read the scan as JSON, with hoserva migrate status --json, each refused disk in the review part of the report carries a refusalCode, shown in the first column.

RefusalWhat it meansWhat to do
The flash and the device name different filesystems (filesystem_mismatch)Unraid's configuration says, for example, btrfs, and the disk reports ext4. Hoserva does not guess which is right.Check the disk in Unraid to find out why they disagree. If you change anything, take a new Flash Backup, then scan again.
The disk is one of several devices of a btrfs filesystem (multi_device_btrfs)The filesystem is not self-contained on this disk.See Unsupported arrays.
The filesystem UUID is also on another disk (duplicate_uuid)Two data disks carry the same ID, which can happen when a disk was cloned from another. Both cannot be adopted.Resolve the duplicate in Unraid, for example by removing a clone from the array, then scan again.
A log was never applied (pending_log)A btrfs disk has a tree log that was never replayed, or an ext4 disk has a journal that needs recovery. Hoserva mounts the disk read-only without replaying anything, so the files in the log would be missing from what it reads. It never replays a log on a disk it does not own yet.Start Unraid and stop the array cleanly, which leaves the log clean. Shut the server down and scan again.
The filesystem could not be shown to be clean (filesystem_unverified)Hoserva could not read the disk's superblock to confirm that it is a single-device btrfs filesystem with a clean log, or an ext4 filesystem with a clean journal. It treats what it cannot confirm as unsafe.Check the disk's cabling and health in Unraid, stop the array cleanly and scan again. If it is refused again, treat it as a damaged disk, as in the next row.
The read-only check failed (integrity_check)The check found a problem in the filesystem. Calculating parity over a damaged filesystem would keep the damage.Repair the filesystem in Unraid, then scan again. Hoserva never repairs a disk it does not own yet.

A disk can also be refused for reasons that do not depend on its filesystem, such as being reported failed or not being read completely. Troubleshooting lists them.

While the verdict is No-go, the import is not offered, so a refused disk has to be resolved and cannot be skipped. Fix the cause, then scan again.

Next steps​