Issue „Can't start hassio_observer:...“„

fisch55 · January 31, 2021, 7:00am

Hi @all,

can anyone help me with this issue… HA is not reachable .

Pi3
Home Assistant OS 5.10
core-2021.1.5
supervisor-2021.01.7


21-01-31 06:48:01 WARNING (MainThread) [supervisor.misc.tasks] Watchdog/Docker found a problem with observer plugin!
21-01-31 06:48:01 INFO (MainThread) [supervisor.plugins.observer] Starting observer plugin
21-01-31 06:48:01 INFO (SyncWorker_3) [supervisor.docker.interface] Cleaning hassio_observer application
21-01-31 06:48:02 WARNING (SyncWorker_3) [supervisor.docker] Can't attach hassio_observer to hassio-network!
21-01-31 06:48:03 ERROR (SyncWorker_3) [supervisor.docker] Can't start hassio_observer: 500 Server Error for http+docker://localhost/v1.40/containers/10288dcc08e49657d7944301482decf0356144cf33dd658c2fb21b6f58a11b46/start: Internal Server Error ("driver failed programming external connectivity on endpoint hassio_observer (dab38835300c62039b2745c0d003cf61df9c21b4a2ba0e32a9f13a5229cc9ac1): Bind for 0.0.0.0:4357 failed: port is already allocated")
21-01-31 06:48:03 ERROR (MainThread) [supervisor.plugins.observer] Can't start observer plugin
21-01-31 06:48:03 ERROR (MainThread) [supervisor.misc.tasks] Observer watchdog reanimation failed!
21-01-31 06:49:03 WARNING (MainThread) [supervisor.misc.tasks] Watchdog/Docker found a problem with observer plugin!
21-01-31 06:49:03 INFO (MainThread) [supervisor.plugins.observer] Starting observer plugin
21-01-31 06:49:03 INFO (SyncWorker_3) [supervisor.docker.interface] Cleaning hassio_observer application
21-01-31 06:49:04 WARNING (SyncWorker_3) [supervisor.docker] Can't attach hassio_observer to hassio-network!

tom_l · January 31, 2021, 7:11am

Connect via SSH and issue the following command:

ha supervisor repair

If that does not help, try this:

ha core rebuild

fisch55 · January 31, 2021, 7:19am

Thx , I will try an give response.

m0wlheld · January 31, 2021, 7:26am

Something on your host is already using a required network port. Check with netstat -tulpen as root user on the host, which process has bound to this port.

fisch55 · January 31, 2021, 7:36am

I will check THX for helping me !

fisch55 · January 31, 2021, 9:52am

Look good !

m0wlheld · January 31, 2021, 10:17am

Indeed. And not a single docker-proxy process, so I’ll assume that you did run netstat while HA was stopped. Which makes me think that HA OS is causing trouble during startup, but as I don’t use HA OS, I am unable to provide further help.

fisch55 · January 31, 2021, 11:01am

ok , which OS I should use?

riddledaxis · April 2, 2021, 4:23pm

I was having these issues as well using HassOS 5.12 in a Proxmox VM and running the latest supervisor (2021.03.6) and core (2021.4.3).

I did the following and it seems to have resolved this error:

stopped HA - Configuration>Server Controls>Server Management>stop

logged in through SSH to my HA VM on the Proxmox server and ran the following commands:

sudo netstat -tnlp | grep :4357 - it showed a tcp6 process listening on this port

sudo fuser -k tcp/4357 - killed that process that was listening

started HA by running ‘ha core start’ and ran ‘ha supervisor repair’ from SSH terminal and once that had finished, I restarted the HA server - Configuration>Server Controls>Server Management> restart

and when the server came back up, the errors in the original post were gone from the supervisor logs - Supervisor>System controls>Log Provider - supervisor and so far have not come back after several server restarts.

Roy75 · April 3, 2021, 5:58pm

are they still gone?

cybermania · April 5, 2021, 8:19pm

This worked for me, but just for a few hours. After a while the errors appeared again.

I started having this after changing RPi IP address to another one on the same network.

Any clue on this?

Thanks,
HS

cybermania · April 5, 2021, 10:03pm

Hi!

Thanks for shedding some light. Is it still OK for you?

In my case (RPi), the command sudo netstat -tnlp | grep :4357, gives this output:
tcp6 0 0 :::4357 :::* LISTEN -

So something is definitely listening (although there is no PID).

For fuser, when I run it gives me this error:

~ $ fuser -k tcp 4357
fuser: can't stat 'tcp': No such file or directory

After some digging I found that in this format it gives no error but still does not close the process that keeps listening still:
‘~ $ sudo fuser -k 4357/tcp’

Any other ideas on this?

Kind regards

riddledaxis · April 6, 2021, 6:25pm

Sorry, can’t believe I missed the slash and the sudo part of the command.

This happened on my server as well.

After 20 minutes or so of me running the ‘sudo netstat -tnlp | grep :4357’ and getting responses saying that something was listening on that port but with no PID.

I decided to go ahead anyway and run ‘ha core start’ and then ‘ha supervisor repair’.

The repair took ages to run, so I just left it and came back to it after 45 minutes or so. The server was running with no error’s in the Supervisor log.

Since, repairing my HA supervisor I have rebooted a lot with updates etc, and these error’s have not returned.

I’m not sure why the errors have come back on your server. The only thing I can think of is did you did stop the HA server before you killed the process?

martusi · April 12, 2021, 5:35pm

Periodically it also happens to me when I update the Observer, I fix it with a restart of docker: “sudo systemctl restart docker” however I have disabled IPV6 on docker and on my Debian host, let’s see if it will do it again.