HA crashed with no obvious reason


I have Home Assistant OS installed on a Raspberry Pi 4 and it has been running flawlessly since several months.

Then, it suddenly crashed with no obvious reason and my only error in the logs is this one:

Logger: async_upnp_client.ssdp
Source: components/ssdp/__init__.py:350
First occurred: 17:14:41 (1 occurrences)
Last logged: 17:14:41

Received error: [Errno 99] Address not available, transport: <_SelectorDatagramTransport fd=25 read=polling write=<idle, bufsize=0>>, socket: <asyncio.TransportSocket fd=25, family=10, type=2, proto=0, laddr=('::', 46583, 0, 0)>

What is the cause for that and how can I prevent it from happening again?

Best regards
Aaron Eisele

Is it this:

It could be this, but what can I do now?
Should I wait for 2023.2 to be released?
This is my production HA install which is too important for me to use a beta version as I want to avoid any unnecessary downtime.
Also, I recently improved the power supply of my RPI4 as I constantly received Undervoltage errors in my logs.

What versions of Home Assistant are you running (Settings → About)?

I’m running version 2024.1.2

You are at the current release of Core but OS is currently at 11.4 so, you should already have the fix that was referenced in GitHub.

  1. How many times has the async_upnp_client.ssdp appeared in your logs?
  2. Did the actual time of the crash match the actual time in the log (17:14:41)?
  3. Were there any changes to your network (the async_upnp_client.ssdp error looks like a network issue)?

I’m more inclined to believe that the crash may have been caused by your under voltage errors - not the async_upnp_client.ssdp error.

  1. As far as I know, it appeared only once but I don’t have the full logs as the Pi and HA were restarted since then (I upgraded the power supply, so no Undervoltage anymore)

  2. I took notes on the crash and from those I saw that it matches the time that I noticed the crash

  3. it may have been caused by my network switch (the one that connects the Pi to the rest of the network) that I think locks itself up sometimes and then needs to be powercycled.