Hi!
We are experiencing UI-related problems. I have no problem using either Firefox (139.0.1) or Chromium (137.0.7151.55) to control pios from my PC (Linux Mint Cinnamon 21.1): I can fully manage the system.
However, I can’t control the pios from my colleagues’ computers, whether they use Ubuntu, Windows, or Mac, and whether they use Firefox, Chrome, or Safari. They can access pioreactor.local, but they can’t control anything. Additionally, the “Inventory” section does not show any information about IPv4, Raspberry Pi, WLAN MAC, or Ethernet MAC.
The problem persists when my PC is offline
Could this problem be related with Tailscale?
Cheers!
This is definitely a MQTT address issue. I’m guessing the leader is on tailscale, your computer is on the same tailscale network, but your colleagues’ computer is on the same (local) network as the leader? Is that correct?
My computer and the colleagues computers, as well as leader, are on the local network. I mean, we all access to the control panel through “pioreactor.local” (not through the IPv4).
I checked the config files and seems to be OK according to the instructions provided to instale Taiscale (which works fine):
shared config.ini
[cluster.topology]
# should be a hostname defined on the network
# See docs: https://docs.pioreactor.com/user-guide/create-cluster
leader_hostname=leader
# something like: leadername.local or a static IPv4 address
leader_address=leader.local
[cluster.addresses]
worker1 = worker1.local
worker2 = worker2.local
# You can add static IPv4 addresses here for other Pioreactors in the cluster. This will bypass mDNS. Example:
# pio01=10.42.0.1
# pio02=10.42.0.3
# The leader's address is automatically added.
# If addresses aren't present, mDNS will be used -> query for <hostname>.local on the network.
[mqtt]
username=pioreactor
password=raspberry
broker_address=100.110.232.30
broker_ws_port=9001
broker_port=1883
ws_protocol=ws
use_tls=0
[ui]
# the UI will be available at the below alias, along with <leader_hostname>.local
# note - it must end in .local, and subdomains are not allowed.
# requires a reboot after changing.
domain_alias=pioreactor.local
# port the webserver is listening to, found in lighttpd.conf
port=80
# protocol the webserver expects.
proto=http
leader config.ini:
[cluster.topology]
leader_address=127.0.0.1
[mqtt]
broker_address=localhost
workers config.ini
[mqtt]
broker_address=leader.local
Also, I get the following error:
Failed to connect to MQTT. Is configuration for mqtt.broker_address correct? Currently set to 100.110.232.30
If tailscale is down (ssh in leader and sudo taiscale down), I can’t retrieve all the UI info even from my computer
Are you seeing this on both computers, or just your colleague’s computer?
Is your colleague’s computer on tailscale? To access the ipv4 100.110.232.30, it should be on tailscale.
Both computers. We finally tried to revert the tailscale changes applied in the configuration files, as well as removing tailscale from raspberry. However, the problem persisted. In the end we performed a fresh installation from scratch.