USB Drive Size, File System Recommendations, and Time Machine on Asus Merlin Routers

[KCi_MK]

While browsing the forums, I noticed some confusion regarding the maximum supported drive size, the best file system to use, and how to set up Time Machine on Asus Merlin routers.

I tested and confirmed the following on the GT-AX11000, GT-AXE16000, and GT-BE98 models:

---

Quick Summary:​

1. Maximum Drive Size:
   • You can use drives of any size, including partitions larger than 4TB.
2. Formatting and Partitioning:
   • Do NOT use the Asus (or Merlin) web interface to format or partition your drive.
   • It limits partitions to 2TB and only supports formatting in FAT, NTFS, and HFS.
   • The recommended file system for speed and data consistency on these Linux-based routers is ext4.
3. Why Use ext4:
   • Specifically developed for Linux, ext4 supports volumes up to 1EB and files up to 16TB.
   • Best performance (lower CPU usage) and more resilient against data corruption than NTFS, FAT, or HFS.
   • NTFS and HFS drivers, provided by Tuxera, are not native to Linux and should be avoided for read-write operations unless necessary.
4. Time Machine Support:
   • Time Machine (Under USB Application->Time Machine) (=AFP; Apple Filing Protocol) works properly, even on the latest macOS 15, if your USB device is formatted correctly for the router’s Linux kernel.
   • However, since AFP is deprecated (starting with macOS Sierra 10.12), Apple now recommends using SMB (version 3.x) for Time Machine backups. Future macOS versions may eventually drop AFP support altogether.

---

How to Prepare a USB Drive for Asus Router (Using ext4)​

I use Fedora Linux for partitioning and formatting, but any modern Linux distribution will work. The router’s kernel doesn’t support all the latest ext4 features, so it’s best to format the partition directly on the router.

Steps:​

1. Connect to the Router Terminal:
   • SSH into your router.
2. Connect the USB Drive:
   • Plug the device into the router’s USB port.
   • It won’t automatically mount due to ext4 incompatibilities.
3. Check the Partition Table:
   • Run:
     

     fdisk -l

   • Ignore the warning:
     

     fdisk: device has more than 2^32 sectors, can't use all of them

   • Find your partition, usually number 1, named something like /dev/sda1.
4. Format sda1 with ext4:
   • Run:
     

     mke2fs -t ext4 -L DEVNAME /dev/sda1

   • Replace DEVNAME with a recognizable label for your drive.
5. Re-mount the Drive:
   • Unplug and reconnect the USB drive.
   • The router should now automatically mount it.

---

​

Optional: Verify and Manually Mount the Drive (Advanced Users)​

1. Monitor System Logs:
   • Keep an eye on the system log during the process to catch any issues.
2. Create Mount-point and Mount the Drive:
   

   mkdir /mnt/DEVNAME
   mount /dev/sda1 /mnt/DEVNAME

3. Unmount and Clean Up:
   • If everything works as expected:

   • umount /dev/sda1
     rm -r /mnt/DEVNAME

   • This prevents potential "ghost" mount points.
4. Re-plug the USB Drive:
   • The router should automatically mount it, and the device will be ready for use.

 

[ColinTaylor]

Also see the Wiki which covers a lot of the same information.

Disk formatting

Third party firmware for Asus routers (newer codebase) - RMerl/asuswrt-merlin.ng

github.com

 

[bennor]

There is what Asus indicates (link below) for; Disk Format, Maximum Partition Size, Maximum Partition Number and File Size, and then there is what people will do in the real world (as evidenced in many past discussions on USB drives, capacities and partition type).
ASUS Router Plug-n-Share Disks Compatibility List

 

[bbunge]

Even so, a router does not make a good NAS. The larger the drive the greater chance for data loss. USB 3 devices can cause issues with 2.4 GHz WIFI. Small thumb drives or small SSD's, yes. Large spinning rust, no!

 

[Tech9]

How to Prepare a USB Drive for Asus Router

There is much shorter method saving future trouble - don't plug the USB drive into this Asus router. When you test it only it will be working, when you actually need the backup data you may find it corrupted. This is not written in the Wiki.

 

[KCi_MK]

A bit more info on 2.4GHz interference with USB 3: https://www.usb.org/sites/default/files/327216.pdf

Agree, spinning HDD's in this configuration is asking for trouble.
SSDs, however, are a different story; they’re becoming larger, more affordable, and often come with high-quality shielded cables to minimize issues.
If substantial storage is needed, investing in a proper NAS is the best approach. That said, what qualifies as "large" storage today is quite different from just a few years ago.

 

[KCi_MK]

MacOS users should generally not worry about Time Machine backup being corrupt.

MacOS performs consistency checks on Time Machine sparsebundle files, as follows:

1. During Backups:
   Time Machine verifies the integrity of individual files as they are written, ensuring that data is correctly copied to the backup. However, it doesn’t check the full integrity of the entire sparsebundle after every backup session.
2. Periodic Verification:
   macOS runs full consistency checks periodicallyor when certain conditions are met, such as:
   • The backup disk hasn’t been used for a while.
   • Time Machine detects potential issues (e.g., sudden disconnection, failed backup attempts).
   • After a forced shutdown or power loss.
3. Manual Verification:
   You can trigger a full consistency check manually using Disk Utility or the Terminal command:
   
   

   sudo tmutil verifychecksums /path/to/your/backup.sparsebundle

   This will verify the integrity of the backup without making changes.
4. Automatic Repair Attempts:
   If macOS detects corruption during a backup, Time Machine will usually attempt to repair the sparsebundle. If the corruption is severe, you might be prompted to start a new backup.

Backups should be stored on at least three different media, with at least one off-site. This approach ensures data redundancy, protects against hardware failures, and safeguards the data from physical threats such as theft, fire or other disasters. There is nothing wrong, if one of those media is an SSD connected to Asus router. It is one of three anyway.
I would not rely on it as the sole backup medium. If it is the only backup option, connecting the USB drive directly to the local workstation hub is much faster and more convenient.
Additionally, sparsebundle files are encrypted, providing strong protection against third-party access—assuming a robust password is used.

 

[JGrana]

As others are warning, USB disks/storage can be problematic on these routers.
IMHO, the main culprit is the lack of a “shutdown -h” command. Only reboot. No good way to safely “power off” these routers when you have USB disks attached other than killing processes and syncing/unmounting before power off - can be a bit tedious.

At least the built in amtm format command offers (and recommends!!!) journaling on ext4.

BTW, ALWAYS use the “service reboot” command. DO NOT use the busybox “reboot” command. I learned this the hard way!

 

[Ripshod]

I find using the good old "halt" command will close everything down in a safe way. Wait 5 minutes then power off.

 

[JGrana]

I find using the good old "halt" command will close everything down in a safe way. Wait 5 minutes then power off.

Ahh, yes, forgot about halt.

@ColinTaylor did mention this in a thread last year:

Last time I checked, issuing the halt command automatically runs services-stop and unmount. But that may have changed so check it on your router.

Us old timers like shutdown.
I’ll create a link to /sbin/halt ;-)

 

[ColinTaylor]

Us old timers like shutdown.
I’ll create a link to /sbin/halt ;-)

Yeah for those of us old enough to remember, it's a BSD vs. Unix System V thing.

 

[sfx2000]

Just saying - TimeMachine support... Not advised as the basics over on the Apple side have changed significantly - moving from AFP/Netatalk to Avahi/Samba2+ along with the transition from HFS+ over to APFS inside the sparesbundle...

Just asking for trouble for data loss...

For TimeMachine support on Macs - best to do direct attached disks, and if network support is needed - find a Mac and setup TimeMachine server support there.

 

[sfx2000]

MacOS users should generally not worry about Time Machine backup being corrupt.

Yes, until one realizes that SparseBundles have a max size before they tend to fall apart - and that is around 100 bands or so, and once exceeded tend to go ker-blam...

TimeMachine on newer Macs uses APFS, and the backups are block based, not file based - and this can lead to problems one slower media, e.g. spinning rust, esp with SMR HDD's and the sessions timing out, normally when TM starts trimming things up, or as they say "Cleaning up"

 

[KCi_MK]

Yes, until one realizes that SparseBundles have a max size before they tend to fall apart - and that is around 100 bands or so, and once exceeded tend to go ker-blam...

I guess that was a typo - 100,000 (100k not 100)?

Performance tend to slow down with too many bands. If slow media is used + network share, it could get messy with time-outs etc.

P.S.: One should use larger band sizes than default 8MB for large sparsbundle files (for example 2TB file with 32MB band size will already have 65,536 bands). For 4TB SSD NAS 16MB band size should be fine, for HDD 32MB...

 

[sfx2000]

I guess that was a typo - 100,000 (100k not 100)?

Performance tend to slow down with too many bands. If slow media is used + network share, it could get messy with time-outs etc.

P.S.: One should use larger band sizes than default 8MB for large sparsbundle files (for example 2TB file with 32MB band size will already have 65,536 bands). For 4TB SSD NAS 16MB band size should be fine, for HDD 32MB...

100K bands... as noted.

General guidance is less than that - 55K seems to be the upper limit across different activities... 55K bands with the default bandsize is around 400gb with defaults on MacOS with the backup on APFS, and the sparsebundle on APFS.

TimeMachine is a rolling window backup over a given timeframe - it's not absolute...

Direct Attach Storage is always the best answer for TimeMachine these days on MacOS... as the storage there is APFS direct on media, and not via a SparseBundle image file...

 

[sfx2000]

Network backed storage - I mentioned direct is preferred, but the second option is network served via another MacOS machine...

 

[KCi_MK]

100K bands... as noted.

The 100,000 bands limit comes from legacy issues with HFS+, where performance and filesystem stability were much more sensitive to the number of files in a directory. Many articles and blogs still recycle this number as a blanket recommendation, but those are outdated assumptions for APFS.

• APFS has no documented hard limit on the number of files in a directory.
• I've seen Time Machine sparse bundles with millions of bands working reliably (especially for long-term backups).
• APFS handles directory lookups in O(log n) time complexity (thanks to B-trees), so performance degradation is much slower than HFS+ (which was closer to O(n) on large directories).

SMB network shares for Time Machine will always use sparse bundle files, regardless of underlying OS. If underlying OS can handle writes reliably, it should not matter. NAS devices are purposely made for storing files, and as such should not have problems with sparse bundle file.

I've seen blogs talking about 500.000 bands with Time Machine sparse bundle using APFS to be safe bet, but I could not find any verifiable data to support any limitation.

The common misunderstanding is confusing performance degradation with filesystem failure:

Scenario	Root Cause	Suggested Fix
Sparse bundle can't mount	SMB/AFP cache corruption	Delete .com.apple.timemachine.diagnostic.loginside the bundle.
Time Machine slow backups	Millions of small bands + Spotlight indexing	Disable Spotlight for the sparse bundle folder (mdutil -i off).
Time Machine backup integrity errors	Power failure during write	Enable journaling or use APFS native snapshots.

At the end, user chooses what and how it will use. Time Machine does provide several checks to ensure integrity. In edge cases, those could be insufficient. Edge cases mostly originate from network glitches or power failures. In case that band is partially written, it will not be rechecked. Design of network is important as is power delivery via UPS for NAS to avoid problems.

Additionally, user can detect corruption in APFS sparse bundle file with:

find YourBackup.sparsebundle/bands -type f -exec sh -c 'shasum "$1"' sh {} \;

If any file cannot be read or hangs, that band is corrupted.