Hi,
I’m wondering if anyone might be able to help me debug an uncommon failure I’ve got with
Balena OS and our ARM board (TM3). It happened approximately once out of 4048 reboots.
Very occasionally we noticed the hostname wasn’t being set correctly by the Balena services, it was set to the kernel compiled hostname, H6.
After looking a bit more into this, I noticed some of the Balena services weren’t running correctly -
root@H6:~# systemctl list-units --failed
UNIT LOAD ACTIVE SUB DESCRIPTION
● balena-hostname-conf.service loaded failed failed balena-hostname.json watcher service
● balena-hostname.service loaded failed failed Balena Hostname Configuration
● balena-engine.socket loaded failed failed Docker Socket for the API
I noticed by doing journalctl -u balena-hostname.service
, I found:
Oct 30 08:18:56 H6 balena-hostname[1387369]: [balena-hostname][INFO] Setting hostname.
Oct 30 08:18:56 H6 balena-hostname[1387369]: [balena-hostname][INFO] Generating default based on short UUID.
Oct 30 08:18:56 H6 balena-hostname[1387369]: [balena-hostname][ERROR] UUID missing from config.json.
Oct 30 08:18:56 H6 systemd[1]: balena-hostname.service: Main process exited, code=exited, status=1/FAILURE
I noticed though ‘/mnt/boot/config.json’ does have a UUID in.
The particular image being used was built with ./balena-yocto-scripts/build/barys -m tm3 -d
Any pointers would be much appreciated!