[BUG] Memory usage not displayed for containers on Raspberry Pi 5 (ARM64)

[MISIKEX]

Description

When running Dockhand on a Raspberry Pi 5 (8 GB RAM, ARM64), the total system memory is correctly detected, but the memory usage for individual containers is not displayed in the UI (shows 0B for all containers).

On a separate server with an Intel i9-12900T CPU, using the same setup, memory usage is displayed correctly for all containers. This suggests the issue may be specific to ARM64 / Raspberry Pi environments.

Environment:
Device: Raspberry Pi 5 (8 GB RAM)
Architecture: ARM64
Docker version: 29.3.0
Dockhand version: latest (fnsys/dockhand)

Expected behavior:
Each container should show its current memory usage in the "Memory" column.

Steps to reproduce

Run Dockhand on a Raspberry Pi 5 (ARM64)
Start multiple containers
Open the Dockhand UI
Check the "Memory" column

Logs

[dockhand] 2026-03-21T07:48:48.864221300Z 2026-03-21T07:48:48.864Z [Auth] Login successful: user=REDACTED provider=local ip=REDACTED mfa=no
[dockhand] 2026-03-21T09:27:52.426824399Z Running as root user
[dockhand] 2026-03-21T09:27:52.427244195Z Docker socket accessible at /var/run/docker.sock
[dockhand] 2026-03-21T09:27:52.476811251Z Detected Docker host hostname: HOSTNAME

============================================================
DATABASE INITIALIZATION
============================================================
     Database: SQLite
     Path: /app/data/db/dockhand.db

[OK] SQLite database opened
     Total migrations: 4
     Applied: 4
     Pending: 0
[OK] Database schema is up to date
[OK] Database initialized (SQLite)

Detected Docker socket at: /var/run/docker.sock
[Crypto] Kernel: 6.12.75+rpt-rpi-2712 (using native crypto)
[Encryption] Using encryption key from /app/data/.encryption_key
[Encryption] Checking for unencrypted credentials...
Hostname for license validation: HOSTNAME

[HostPath] Detected container ID: c9e9bade2816
[SubprocessManager] Starting Go worker (/app/bin/collection-worker)...
[Scheduler] Starting scheduler service...
[collector] starting...

[Scheduler] System schedule cleanup: 0 3 * * * [Europe/Budapest]
[Scheduler] System event cleanup: 30 3 * * * [Europe/Budapest]
[Scheduler] Volume helper cleanup: every 30 minutes [Europe/Budapest]
[Scheduler] Refreshing all schedules...
[Scheduler] Registered 0 container schedules, 0 git stack schedules, 0 env update check schedules, 0 image prune schedules
[Scheduler] Service started

Listening on http://0.0.0.0:3000/ with WebSocket
[SubprocessManager] Go worker ready

[collector] configured env 1 (HOSTNAME) type=socket base=http://localhost
[collector] metrics interval set to 30000ms
[collector] event mode=stream pollInterval=60000ms

[HostPath] Cached 2 mount(s)
[HostPath] Detected own network: bridge-pro
[HostPath] Detected host path for /app/data: /var/lib/docker/volumes/dockhand_data/_data
[Startup] Host data directory detected: /var/lib/docker/volumes/dockhand_data/_data

[SubprocessManager] Environment "HOSTNAME" (1) is now online

Browser logs

Dockhand version

1.0.21 - d256fe9 (Mar 13, 2026)

Hawser version (if used)

No response

Connection mode

Shocket

Please confirm the following

• I have already searched for relevant existing issues and discussions before opening this report.
• I have updated the title field above with a concise description.

[Akshunhiro]

Have you added cgroup_enable=memory & cgroup_memory=1 into /boot/firmware/cmdline.txt?

[MISIKEX]

Have you added cgroup_enable=memory & cgroup_memory=1 into /boot/firmware/cmdline.txt?

Dear Akshunhiro,

Yesterday I migrated Docker from my Raspberry Pi to an Intel-based server. Unfortunately, I can’t verify whether it would have solved the issue, but if I set up Docker on a Raspberry Pi again in the near future, I’ll definitely check it.

I’m sure that I hadn’t configured this setting before, so I do believe it might have resolved the problem—but as I mentioned, I’m not able to confirm that right now.

Thank you very much for your quick response!

Wishing you perseverance and a relaxing weekend… :)

[radoeka]

Thanks for reporting the issue, I wanted to report the exact same issue.

Have you added cgroup_enable=memory & cgroup_memory=1 into /boot/firmware/cmdline.txt?

I'm still on Raspberry PI5 and this made it work!
The same procedure must be used in case Hawser is used for remote PIs.

As a side note, the memory metrics became visible in Komodo as well once the cgroup_memory setting is enabled on the PI

For me the ticket can be closed, solution provided.

Detailed instructions:

Display Memory Metrics

For Dockhand Hawser on Raspberry PI5 cgroup_memory must be enabled Dockhand issue 804 to display memory metrics.

Add the following settings: cgroup_enable=memory cgroup_memory=1 at the end of the line in /boot/firmware/cmdline.txt. Do not add a new line, keep only 1 line in that file.

Verify the modification

Count the number of lines in the file:

wc -l /boot/firmware/cmdline.txt
grep cgroup_enable /boot/firmware/cmdline.txt
grep cgroup_memory /boot/firmware/cmdline.txt

The result should be as follows (example):

1
console=tty1 root=PARTUUID=39xxxxc6-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=EU cgroup_enable=memory cgroup_memory=1
console=tty1 root=PARTUUID=39xxxxc6-02 rootfstype=ext4 fsck.repair=yes rootwait quiet splash plymouth.ignore-serial-consoles cfg80211.ieee80211_regdom=EU cgroup_enable=memory cgroup_memory=1

Reboot

To activate the modification reboot the system: reboot.

Verification

After reboot, check with cat /sys/fs/cgroup/cgroup.controllers or use grep memory /sys/fs/cgroup/cgroup.controllers that the memory cgroup is enabled.

The result should be: cpuset cpu io memory pids.