Device ended up in a loop restarting services after new release

Product support
1

After deploying a new release (4 containers), one (out of ~10) devices (RPI4) ended up in a retry loop:

  1. Starting services → logs from some of the containers appear.
  2. A few minutes later it fails to “wrap up” the release (see error below)
  3. The loop starts over

…
2025-10-13T12:00:12+02:00 Deleting image ‘registry2.balena-cloud.com/v2/a5…’
2025-10-13T12:00:12+02:00 Deleting image ‘registry2.balena-cloud.com/v2/15…’
2025-10-13T12:00:15+02:00 Failed to delete image ‘registry2.balena-cloud.com/v2/a5…’ due to ‘socket hang up’
2025-10-13T12:00:15+02:00 Failed to delete image ‘registry2.balena-cloud.com/v2/15…’ due to ‘socket hang up’
2025-10-13T12:00:12+02:00 Supervisor starting
…

BalenaOS: 6.5.50+rev2

Rebooting device fails and ssh isn’t available.

How one can recover from this?

2

Hi Carloz,

Thanks for sharing.

Do you have shell access to the device, and can you use the balena engine on the HostOS to manually stop the services?

3

I do not.

Trying to open the terminal just goes into loop
SSH reconnecting…
SSH session disconnected

Rebooting the gateway (via UI) results in {“Data”:“”,“Error”:“read ECONNRESET”}

Restarting all services (via UI) results in {“status”:“failed”,“message”:“connect ECONNREFUSED /var/run/balena-engine.sock”}