Got a problem yesterday - my all zigbee/mqqt device are no longer available after front end went unresponsive (which was quite frequent for me, like once a day at least for last few months) - usually rebooting the box (i have a thinclient, with linux arch and Home Assistant core installed on it).
But this time it went down for good, whatâs worse i cannot any longer connect to broker when using âhttp://192.168.1.69:8081/#/â - connection is refused. The frontend of the HA is accessible and other integrations beside MQQT are working.
I am no linux or HA power user so definitely needy your help to troubleshoot this thing
So far i noticed in hostâs console after typing âtopâ abnormality that command from user zigbee2+ consumes around 120% of CPUs time as shown in screenshot attached to this post.
There has been no direct change to system made - it was going unresposive like thet from time to that - i did run âpacman -Syuâ on a box but that was day before - and box was rebooted during that time - so i hardly believe update to system made any cause.
oh, youâre going to love the answer - a buddy of mine did it and kinda all the specs are gone with him now because heâs eternal douche
I can do all the troubleshooting you want me to - but i need hand holding here and thatâs the problem
edit; the alternative here is to extract the profile from this instance and later IF needed i can set up HAOS on my own etc - but thing is i have a tons of config there and would like to salvage or bring this one back to life.
In your first post you said that Home Assistant is still accessible. In that case, I recommend you make a full backup (Settings â System â Backup). Save the backup file and restore it on a new installation of Home Assistant OS.
HA OS is the preferred installation method because thereâs far less maintenence involved and has simpler upgrades (compared to what you currently have). The MQTT Broker is available as an Add-on for HA OS.
Okay, cool - is that back from Core version will be compatible with new one?
However - are there any steps i can do (including command lines) for me to maybe reinstall the mosquito or reconfigure it? Any other ideas besides tearing this whole thing down?
any other ideas guys? The issue here clearly loos like connection got broken for some reason - i got spare Zigbee dongle just in case - i got full access to machine via putty - any ideas how to troubleshoot this?
Home Assistant and Zigbee can be installed in maybe a half dozen or more different ways.
Your friend appears to have configured for you what is probably the most complicated and most difficult method to maintain. Usually this method is only used by seasoned Linux (and/or Python) experts.
Your long-term goal should be to scrap this installation and migrate to one that is easier to maintain. The recommendation (above) is to install Home Assistant Operating System (HAOS) which offers lots of tools to help inexperienced users.
If you can backup your existing HA you might be able to restore the backup onto the new HAOS which would minimize disruption. There is a backup feature in the HA web interface under settings â system.
Now, as far as repairing this installation â lots could be wrong here. If a reboot isnât working, then a simple process restart probably wonât fix either. First run df -h to get a report of your remaining storage to ensure none of the Linux volumes have filled up.
The âbrokerâ you were connecting to on the :8080 url (oops, see below) is actually the zigbee2mqtt web frontend, a service which translates between the Zigbee protocol and mqtt messaging over IP (likely truncated âzigbee2+â on your top screenshot). You are correct in deducing this should not be using 123% cpu. Youâll need to review the logs to find out why the app is malfunctioning. If the :8080 console comes back, you may be able to view the logs there; if not, you youâll need to find the directory where itâs storing log files â I donât have enough info to help you there. The Z2M project has a lengthy list of startup troubleshooting topics here.
Edit: just saw youâre using a url with :8081 which could be something else? Even Mosquittoâs web console (rarely used) defaults to :8088 so itâs possible your friend changed some defaults.
nothing concerning here - although i have been delaying with some issues in the past with some log messages depleting space
as for the ports thing - yea i know that in order to acess my HA front end i used 192.168.1.69:8123 while it was 192.168.1.69:8081 for the the mqqt interface
Maybe thet can give you some clue or you got some idea to follow? In the meantim Iâll review that troubleshooting topics as well.
btw; that exact moment when this all went back was the reboot of the box - do you think mqqt config file could became corrupted or maybe overwritten so that thereâe no communication there?
mqtt or zigbee2mqtt? They are different services, neither of which usually uses :8081 for webpage. The Zigbee frontend looks like this with a list of your devices and a map of connections, but defaults to port :8080. The MQTT broker, which could be Mosquitto or something else entirely, rarely has a web frontend (though notably the bevywise broker has a GUI on :8080 just like Z2M which could explain why one was moved to :8081). Step one is to figure out which service is actually failing.
Apologies - based on your screenshot we can finally determine that i am running indeed the Zigbee2MQTT - however i have seen mosquito in top at various times.
As for the reason for service failing - would it matter i was experiencing regular issue (every 24hrs) when Zigbee devices became unresponsive? It was strange but i had it for months really. I am wondering - maybe there is (and was) some hardware issue going on with the dongle?
Now, i mentioned i bought spare Dongle⌠the make and model is Sonoff ZBDongle Plus-E
And this how this devices presents in my âdevicesâ :
I am not familiar with this add-in - you think i can take any leverage of that here?
So again just the output from top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
