mDNS and systemd-resolved high CPU usage

[eatoff8]

Hardware Environment

VirtualBox/VMWare player (both gave same symptoms)

Home Assistant OS release:

• Fresh installation of release 5.13, restore from backup 5.13 same issue

System Health

version	core-2021.6.3
installation_type	Home Assistant OS
dev	false
hassio	true
docker	true
virtualenv	false
python_version	3.8.9
os_name	Linux
os_version	5.4.109
arch	x86_64
timezone	Australia/Perth

Home Assistant Community Store

GitHub API	ok
Github API Calls Remaining	4925
Installed Version	1.12.4
Stage	running
Available Repositories	824
Installed Repositories	12

Home Assistant Cloud

logged_in	true
subscription_expiration	14 June 2021, 8:00
relayer_connected	true
remote_enabled	true
remote_connected	true
alexa_enabled	false
google_enabled	true
can_reach_cert_server	ok
can_reach_cloud_auth	ok
can_reach_cloud	ok

Home Assistant Supervisor

host_os	Home Assistant OS 5.13
update_channel	stable
supervisor_version	supervisor-2021.05.4
docker_version	19.03.15
disk_total	30.8 GB
disk_used	5.5 GB
healthy	true
supported	true
board	ova
supervisor_api	ok
version_api	ok
installed_addons	Samba share (9.5.0), File editor (5.3.1), Home Assistant Google Drive Backup (0.104.3), Mosquitto broker (6.0.1), AdGuard Home (4.1.6), Glances (0.12.1), DOODS (2), Visual Studio Code (3.4.1)

Lovelace

dashboards	1
resources	7
views	18
mode	storage

Description of problem:
High CPU utilisation from systemd-resolved. CPU usage slowly creeps up and after about 20 mins is pegged at 100%. This happened with a fresh install (no backup image used, jsut a generic user created) and when i tried restoring from a backup. I have traced it back to running mDNS on my unifi system. If I disable mDNS the CPU usage of systemd-resolved drops back to under 1%. BUt without mDNS i cannot push my chromecast devices onto another VLAN.

Here is an example of a fresh install - https://imgur.com/sWLEeuf
Here is an example of my restored from backup - https://imgur.com/undefined

[agners]

When checking journalctl, do you see messages like described in #1163?

Just to understand the architecture a bit better: Is your Home Assistant installation dual-homed (e.g. in the Chromecast VLAN as well as in a regular LAN?)

[agners]

Also: Do you have maybe test installations running with the same Hostname or Instance name?

[eatoff8]

The home assistant install is not dual homed, just on the main LAN. I only ever had one instance running at a time. I tried renaming the host but didn't seem to make a difference.

I'll check the kernel logs tonight

[eatoff8]

When checking journalctl, do you see messages like described in #1163?

Yes, I see exactly what is in that issue. For whatever reason i cant copy the text from the window, but here is a screenshot - https://imgur.com/a/vTsPxFa

No other instances of home assistant are running on my network.

[agners]

Can you try changing the hostname to something other than homeassistant? You can do that in the System tab of the Supervisor section.

[eatoff8]

Can you try changing the hostname to something other than homeassistant? You can do that in the System tab of the Supervisor section.

I renamed it to "HA" which fixed it. But then a restart of the VM brought it back.

[agners]

I am guessing this is the mDNS feature you are using? https://help.ui.com/hc/en-us/articles/360001004034-UniFi-Best-Practices-for-Managing-AirPlay-Chromecast-on-UniFi-Network#enable

From what understand this really plays back the same mDNS packets to Home Assistant, and it seems that systemd-resolved detects that as a hostname conflict. Can you try follow this advice and disable mDNS reflector (while keeping mDNS repeater active)?

#1163 (comment)

In any case, since this really seems to be the same issue as #1163 I am closing this. Please post your findings in #1163.