Preconditions

• Robot is ON, PC/Steam Deck connected to the unit's WiFi AP.
• SW-5 has passed (release image loaded / "Switch ARU" done).
• SSH access to the control PC (root@10.42.0.106).

Steps

1. Disk space & filesystem health

ssh root@10.42.0.106
df -h /                                     # root filesystem usage
mount | grep ' / '                          # confirm rw, not remounted read-only
dmesg | grep -iE 'I/O error|read-only|ext4|corrupt' | tail

2. Containerd store size & disk backing it

du -sh /var/lib/containerd      # total size of the image/container store right now
crictl images | wc -l           # how many images are present right now
df -h /var/lib/containerd        # free space + % used on the disk backing that store

3. Running image matches the deployed tag's registry digest

export CONTAINER_RUNTIME_ENDPOINT=unix:///run/k3s/containerd/containerd.sock
C=$(crictl ps -q --name release-aru | head -1)
crictl inspect $C | grep imageRef               # local digest, e.g. @sha256:...
crictl images --digests | grep <deployed-tag>   # registry digest for same tag
# the two must match — a mismatch = stale image under a reused tag

4. Expected payloads present in the image

crictl exec $C ls /firmware      # expect: actuators  modules  elec_stacks

Expected

• Disk: root and the disk backing /var/lib/containerd mounted read-write;

no I/O / read-only / corruption errors in dmesg; ≥ 100 GB free on that disk.

• Store: du / crictl images return without error; no more than 5 images

present (crictl images -q | wc -l ≤ 5) — no runaway accumulation of stale images.

• Image state: the running image's digest matches the registry digest for its

deployed tag (no stale image under a reused tag).

• Payloads: actuators, modules, elec_stacks all present under /firmware.