PuppyIP Resource Center
Network & Connection Troubleshooting 10 min Published 2026-11-17

KWRT Boots the Wrong Root from USB and NVMe: Check Duplicate PARTUUIDs and F2FS Growth

With a USB boot drive and a cloned NVMe attached, duplicate PARTUUIDs can make root selection ambiguous. Confirm the active root before changing identifiers or capacity.

KWRT x86 Duplicate PARTUUID USB boot NVMe F2FS OpenWrt storage

Service eligibility and regional restrictions

PuppyIP serves only compliant overseas businesses and their authorized personnel. Proxy services are not available in mainland China. The service may only be used for lawful business activities outside mainland China. Use of this service within mainland China is prohibited.

Hosting a proxy IP or server overseas does not change these restrictions. The service must not be provided to end users in mainland China through relaying, forwarding, sharing or resale. Before use, read the Terms of Service.

Key Takeaways

  • PARTUUID identifies a partition-table entry; it is different from the F2FS filesystem UUID and device path.
  • Use /proc/cmdline, findmnt and block-device data to identify the active root; do not infer it from a drive label.
  • Changing a partition GUID or boot parameter requires a backup, a known BIOS/bootloader path and recovery media; a wrong value can prevent boot.
  • Partition growth and F2FS filesystem growth are separate steps. Choose tools only after confirming the root device and layout.

Separate PARTUUID, filesystem UUID and the root device

If an x86 router has both a USB boot device and a cloned NVMe attached, duplicated PARTUUIDs can make it unclear which root partition a boot parameter selects. The mounted device also depends on firmware boot order, kernel command line, partition enumeration and overlay configuration; seeing USB or NVMe in a disk list does not prove which one supplied root.

PARTUUID identifies a partition-table entry; a filesystem UUID is stored by the filesystem itself. Changing a filesystem UUID may not fix a duplicate PARTUUID. First collect device, partition, mount source and boot-parameter data, then decide whether to change an identifier, the bootloader target or storage size.

Read-only checks to identify the active root

From a local console or an existing SSH session, read the kernel command line, mount source and block-device identifiers. Compare the root= parameter in /proc/cmdline with findmnt, lsblk and blkid. Identify root, boot and overlay/data partitions separately. If a utility does not display PARTUUID, use an already-installed read-only partition tool; do not install or modify storage just for this check.

Save the output and confirm how devices map to slots before writing anything. Do not publish real serial numbers, WWNs, MAC addresses or private device paths in an article, screenshot or ticket. The commands below only read current device state.

Read boot parameters, mounts and disk identifiers

sh
cat /proc/cmdline
findmnt -no SOURCE,FSTYPE,OPTIONS /
lsblk -o NAME,TYPE,FSTYPE,SIZE,PARTUUID,UUID,MOUNTPOINTS
blkid
From a local console, read the active root, filesystem and partition identifiers. Remove hardware serials, WWNs, real mount paths and other device identifiers before sharing.

Handle duplicate PARTUUIDs and the boot target safely

If two disks really have the same PARTUUID, identify whether BIOS, GRUB, Syslinux, extlinux or device firmware manages the boot chain. Find the root= value the kernel actually reads. After selecting the disk that should remain the boot target, either generate a unique partition GUID for the other disk or update the bootloader reference; these changes must agree. Changing a partition GUID affects every boot entry that references the old value.

Before writing, create a full-disk or partition image, back up boot configuration and prepare bootable recovery media. Use the relevant partition tool only after confirming the device model, partition-table type, target partition and boot configuration. There is no safe universal command for every partition table and boot manager. If the active bootloader is unknown or the new identifier has not been written into its configuration, stop and have the image maintainer verify it.

Verify partition growth and F2FS growth separately

A cloned partition may retain its original small size even when the NVMe is larger; a filesystem does not necessarily use new disk capacity automatically. First inspect the block-device layout to confirm which partition contains root and whether unallocated space follows it. Then use a partition-table tool appropriate to that layout to extend the boundary. This changes disk metadata, so keep a recoverable backup and verify the target before applying it.

Only after the partition has grown should you extend the filesystem using a method supported by the running kernel, F2FS tools and mount state. Do not apply F2FS operations to ext4 or a read-only SquashFS root, and do not format, repair or resize before confirming the target partition. Check partition capacity and df filesystem capacity separately.

Check boot selection and usable capacity after a restart

If identifiers or partitions were changed, verify from a recovery environment that GUIDs are unique, the bootloader points to the intended root, boot settings were saved and the backup is readable. Then reboot during a local maintenance window and recheck /proc/cmdline, root mount source, overlay source, partition identifiers and df. Confirm that both disks remain distinguishable and that the system did not unexpectedly boot from the old USB device.

Record startup stability, free root space and router service state separately. A larger partition in lsblk does not mean the F2FS filesystem expanded; a changed df output does not prove the next reboot will select the intended disk.

Rollback and device-specific limits

If root fails to mount after reboot, use recovery media to restore the partition-table and bootloader backup. Do not keep trying different partition IDs on the data disk. After changing a GUID, every boot entry that referenced it may need an update. Avoid automatic repair or reformatting until filesystem-metadata risk is understood.

The exact device, BIOS/bootloader, partition table, overlay layout and OpenWrt/KWRT image affect the procedure. MIPS devices, different x86 images and external-drive layouts do not share one universal operation. If the active boot source is unclear, keep the read-only findings and stop. Any root resize or identifier change needs local recovery capability.

Sources

Frequently Asked Questions

Are PARTUUID and the F2FS UUID the same thing?

No. PARTUUID identifies a partition-table entry; UUID identifies a filesystem. Check which value the boot parameter references.

Will root expand automatically because the NVMe is larger than the USB drive?

Not necessarily. A cloned partition may retain its original size. Expand the partition and filesystem separately, then verify with df.

Can I change PARTUUID and reboot right away?

Do not change only the identifier. Confirm whether the bootloader and root= parameter use the old value, and prepare a backup and local recovery media.

How can I tell which disk the running system booted from?

Compare /proc/cmdline, findmnt, lsblk and blkid mount sources and partition identifiers. Do not rely on a device label alone.