Unreliable Reboot of Hass.io on Raspberry PI 3

If I use this CLI command hassio ho shutdown to safely shutdown and then cycle the power on my Raspberry PI, Hass.io does not restart.
In most cases there is no Ethernet connection to hassio. Sometimes an Ethernet connection is made to hassio but it becomes inactive about a minute later.
My solution is to cycle the power on my Raspberry PI again. This time it is a hard shutdown since I have no SSH connection, via Ethernet, to run the above CLI command. The Ethernet connection to hassio comes up in about 30 seconds. HomeAssistant UI starts about 2 minutes later.

Anyone else seeing this? Or is my Raspberry PI defective in some way?

Also see my related issue when the router is slow to reboot.

I would suggest stopping HA first, then shutdown or restart the host.

I was experiencing this also for a while, I ended up switching to a faster SD card, and turning of my Unifi Controller add-on I was running (moved it to a container in my NAS), and I haven’t had the problem in a while. Unfortunately I never found a smoking gun so I just did a bunch of things and haven’t had the problem where it doesn’t even have Ethernet when it turns back on. As with you the only thing I could do was cycle the power a second time.

I tried:
hassio ha stop
hassio ho shutdown
then cycle the Raspberry PI power.
I get a hassio Ethernet connection in about 30 seconds (good) but the Ethernet connection is dropped about a minute later.(bad)
My only solution is to cycle the power again.

You would think that hassio ho shutdown would include hassio ha stop as a subset. If not, I would call that a bug.

I am using a Samsung 32 EVO HC I, microSDHC card. I saw that it was recommended for use in a Raspberry PI 3.
So far as I can tell the OS on the card has not been corrupted even after many hard reboot cycles.
I see the HA database (home-assistant_v2.db) being unfinished and rebuilt using temporary home-assistant_v2.db-shm and home-assistant_v2.db-wal files. HA seems to recover from hard reboots but not from a soft reboot, ie you did a controlled shutdown.

I also had this problem and swapped my SD card to resolve, the old one was too slow. Are you sure your EVO is genuine?

It could also be a faulty pi or power supply.

This posting seems to relate to my reboot issue after a soft shutdown.
It describes a race condition found in an Ethernet driver for a rasp pi 3+ (mine is rasp pi 3)
I built my HA back at hassio 0.76. Maybe the latest image (1.13) fixes this driver issue?
I will give it a try.

Interesting aside. I have found that rebooting by shorting the rasp pi 3 RUN pin to ground is reliable.

Just curious. Are you running this on resinOS or hassOS?

This is a HassOS issue. Seems to be a race condition in the Ethernet driver on a reboot.

Just curious why you are power-cycling the Pi?

Its always possible that there is a kernel/driver issue. I’m on 1.13 and have not noticed this.

That is what happens when there is a power outage.
I am now trying a complete rebuild using the latest hassos image for rasp pi.

One of the main reasons for me to switch over to docker on a Linux mint OS , sometimes it just doesn’t boot back after a powercut

SOLVED!!!
I rebuilt my HA system using the latest hassio image here.
Not too difficult if you have a recent snapshot saved on your Windows 10 PC.
The HassOS image version shows up under Hass.io -> SYSTEM

My theory is that the image I was using had a defect in the Ethernet driver.

I would encourage everyone to use the hassos 1.13 image for their Raspberry Pi 3

1 Like