Hi
I have some recurring problems since more than one year with both HA updates and HAOS updates. I run HA on a dedicated NUC with HAOS.
Most of time when I want to do an HA update, it states it’s impossible as no Internet access (although it has) and similar error for HAOS. Network config is nothing special, local IP in IPv4 with NAT in my router, and IPv6 direct.
Now today got an new error when trying to update HAOS from 18.2 to 18.3:
Failed to perform the action update/install. Error updating Home Assistant Operating System: File descriptor 17 is used by transport <_SelectorSocketTransport fd=17 read=polling write=<idle, bufsize=0>>
How to fix these issues one time for good ? It’s the only HAOS install among multiples I manage that have that issue and can’t find anything here or on HACF on how to fix it !
Only way so far since a year to succeed to do upgrades is to do a hard reboot of machine (the software reboot by HA never works, it hangs during shutdown process if I remember right.
It’s not hardware related, is it?
You have at least 2gb of memory or more and a bunch more drive space available than used drive space (2 to 1 is ideal)
No hardware problem related, HAOS is running on a dedicated beefy NUC. Out of the problem of updates impossible, everything else is working great !
It has 8Gb of RAM and 34% of storage used (75Gb used on total of 220Gb).
According to Google your error might be related to integrations or apps that has many websockets open, “leaking” file descriptors. A full host reboot should clear it up.
oki but problem is there since more than a year and it worsens more and more ! Now only way is to do a hard reboot by cutting power and then I might have a chance to do one update before it blocks again
An easy way to find the problematic integration or app ? out of disabling all of them to find the culprit one
Well I have narrowed down the problem, it’s clearly an official integration that is the issue. I started first to disable all custom extensions thinking they have a higher rate of bad programming than official ones but it didn’t change anything.
I ended up disabling all integration out of few ones system and then problem gone ! I’m still very confused how an integration can mess up so much the core HA itself preventing it to do any updates or reboot…
Will reactivate few ones and try to find the culprit…
Update: problem of reboot by itself impossible is still there so I guess only solution will be to do a clean reinstall of the whole thing as it looks messed up beyond recovery
I don’t know so far and considering quantity of integrations that might be concerned I’m not going to loose time to try to find which one ! I’m going to reset the whole thing and reinstall it, that problem has pissed me off since at least a year and without any easy solution to find what/who is the culprit
I had checked in past logs but they are so much noises there it’s impossible to find anything interesting
I still don’t understand how an integration can mess up HAOS itself at such point, it looks like a serious issue regarding reliability
