Replies: 1 comment
-
|
I'm now experiencing shutdowns using Raspbian with |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I have HAOS running on a Raspberry Pi 5 with an official 27W PD 3.1 5V5A power supply, off a USB-SATA SSD. This was an upgrade from a two-year old installation on an RPi4, using a fresh install on RPi5 and a restore from backup. It worked perfectly for a month, without any of the reboot-related issues that were reported here in December.
Yesterday morning I found Home Assistant offline. The RPi5's power LED was red and it didn't respond to the power button. I had to disconnect power to make it boot up again, and everything was fine for the rest of the day. Around midnight, ~15 hours later, it died again. I cycled power again, and this time it powered off within minutes.
I plugged in a monitor and keyboard to see what was going on. The bootup sequence was normal and went into the CLI, where I could type in commands.
host logssuggested HA wasn't fully up — Docker containers were still being started up. About a minute later the display signal cut off, and then the RPi5's power LED changed from green to red. Power button still didn't initiate bootup and needed physical power cycling.The only thing I had changed recently was to add an RTC battery, so I removed it. No change in behaviour though — HAOS still booted up normally, opened into CLI, and then powered off.
To confirm it wasn't a hardware fault, I removed the HAOS SSD and inserted an SD card with Raspbian. This booted up normally and didn't power down. When I replugged the SSD (USB), some of the partitions auto-mounted and I could look around the data volume. Unfortunately, this doesn't appear to contain system logs.
Next, I used a debug probe in the UART port to log output:
At this point the output stopped until HAOS booted up into CLI. When it shutdown again, this appeared:
The garbled output looks like a change in the baud rate. I'm using a Waveshare RP2040-GEEK with their
picoprobefirmware. I don't have an official Raspberry Pi debug probe, but I'll get one now.The time to shutdown seemed to be getting shorter with each power cycle, so I unplugged everything and left it overnight. This morning it powered up and stayed on for several minutes, so I opened HA's web dashboard, went into the Scrutiny add-on's UI and checked the SSD's SMART data. Everything was normal. Since there is nothing so far indicating a hardware problem, I went into system settings to check the logs for whatever errant add-on or integration is causing shutdowns… but HAOS had shutdown again. This time it lasted maybe 15 minutes.
So far I have:
ha host logsanddmesgdon't have streaming output, so I can't see what happens at the moment of shutdown.How do I debug this? Which log files should I be looking at, and which partition and path are they on?
Update, one week later:
HAOS on the RPi5 is still powering off in a few minutes, but I've observed some things:
After 24+ hours powered off, I got 55 minutes uptime. An immediate power cycle after shutdown results in an uptime as low as 30 seconds. This feels like it's related to a capacitor discharge, or maybe thermal cutoff.
It's not the power supply. I moved the SSD to a powered USB hub and tried replacing the official PD 3.1 charger with a Radxa 3.1 charger — which also supplies 5V5A, but which I previously found unreliable unless using a powered USB hub. There's no difference with any of these power arrangements.
I reflashed the SSD with HAOS 11.5 for RPi5 (marked as "Preview" in Raspberry Pi Imager) and tried restoring from backup. It powered off during restore — first time after restore started, second time when uploading the tar file itself.
At this point I suspect thermal cut-off. I had to reinstall the active cooler when swapping the case and it may not be seated properly. I'll check on that and update here.
Beta Was this translation helpful? Give feedback.
All reactions