I had a few of my operators complain they couldn't connect to our Ignition project on their phones when out in the field (via web browser) and I wrongly assumed they were in poor reception areas when encountering this. This week while traveling out of state, I noticed I could not connect to our project over cellular either - I can however connect to the gateway page. If I switch to WiFi the project loads without an issue. On 5G I have around a 500mpbs down 10mbps connection up so it's definitely not a coverage issue.
On the gateway I see an unauthenticated/unauthorized perspective project session for my phone when attempting to connect or when leaving the tab open. If I attempt to connect I see a StatusRoutes log with the message: Requested session for "xxxxx", does not exist. (where xxxxx is my project name, screenshot below). Launching a new session in incognito or terminating the session from the gateway and trying again yields the same results.
After entering credentials it gets to 'Connecting...' then after a few seconds it rolls back and hovers over no text before kicking back to 'Continue to Log In'. This behaves the same with the iPhone app and Android app.
Some information on this system,
project and gateway (v8.1.30) accessed via HTTPs on our hosted domain
SSL/TLS port isn't firewalled (not using default port 443)
Ignition host installed as docker container (Linux VM)
affects **every cellular connection (I thought initially it was only some, but that doesn't appear to be the case)
**edit- this only appears to affect T-Mobile (not Verizon or AT&T), see more details below.
If anyone has any ideas I'd appreciate the feedback.
Inspecting my phone's Chrome session shows a Websocket connection error - HTTP authentication failure.
I'm confused by the HTTP authentication failure since this is an HTTPS connection.
These status log messages seem to indicate that someone was logged into the Gateway Web Interface > Status > Connections > Perspective Sessions page as user "aaron" and clicked on the More > Terminate button next to whatever Perspective client sessions they decided to terminate (see docs). I'm not sure if it is related to your web socket issue.
Do you have any middleware between your browser and the Gateway that we should know about? Things like Web Application Firewalls, Load Balancers, Reverse Proxies, etc? It's possible that something is interferring with the connection somehow.
These status log messages seem to indicate that someone was logged into the Gateway Web Interface > Status > Connections > Perspective Sessions page as user "aaron" and clicked on the More > Terminate button next to whatever Perspective client sessions they decided to terminate
Correct - for that screenshot I had terminated the instance in order to refresh the StatusRoute log.
No middleware, I have only firewalled (via AWS) a few microservices inside that same EC2 instance while leaving the Ignition Gateway totally open to the internet.
Setting the Perspective Permissions to 'Public' (Project Properties->Perspective->Permissions) is a work around, but beyond for troubleshooting purposes it isn't a viable solution.
Hm... This affects our users in Florida, East Texas, and Central Texas. Are you using a secure connection / domain to connect to your perspective project?
I've got a big ask of you - any chance you know anyone with an Android 14 OS to launch your project on? Bonus would be if they're on TMobile. I'm on a Pixel 8 running Android 14
Thanks Paul, this issue only affects projects that are using authentication - the demo you referenced is seemingly public.
I mentioned above, setting the Perspective Permissions to 'Public' (Project Properties->Perspective->Permissions) is a work around, but beyond for troubleshooting purposes it isn't a viable solution.
Any clue on how you solved this in the end? I have a similar scenario with a project behind Azure Proxy Connector. The backend page works well but projects get stuck in ‘Connecting’. Everything points to a problem with WebSockets.
Unfortunately not - reviewing my support ticket notes, we exhausted quite a few tests without a conclusion. However, things did start suddenly working over time and we haven't had this issue pop up, or at least reported to me, in a year now. Sorry