Title says it all. I use Node-Red to make all my automations in HA. This works nice, I find it visual very easy to create and adjust stuff.
Since a while, this is going bad. My automations work, but not very stable. If I go into Node-Red I can see everything, sometimes values change, but not stable. I can’t however change anything. If I want to change devices for example, the lists are empty, on new nodes or old ones that I try to change. I can no longer push the little square buttons on inject nodes, debug is empty and the logs seem to not work.
Everything visualy tells me Node-Red is dead, wierd thing is, all my flows run. I have some set up behind a button, they work. Other way around, I have some flows setting sensors with data, they don’t all work… Very wierd and realy grinding my gears here, help!
This sounds like the known issue on the HA addon for node red (eg. here). If you can use node red fine by accessing it via port 1880 that will probably be proof (you may need to enable the port in the config).
Try updating node-red-contrib-home-assistant-websocket to 0.74.2. You may need to use :1880 to do so. That fixed it for me.
I wouldn’t hold your breath - there’s been similar open issues for quite a while. I’m pretty sure I had to do some editing of entities to get it fully working as well. If you hunt around in the issues you’ll find it I think.
I have the NR addon for HA, but I don’t actually use it - I only have it as a sandbox. I have a separate NR and I’ve not updated the HA addon for NR for ages because the service->action breaking change is going to take me too long to make it work (I have a lot of NR code setting payload.service and relying on domain being set in the node and this is all gone). So you could try downgrading to version 0.65.1, although I’m not sure how easy that is from within the NR addon.
Or give up and access via :1880 until the issues get closed.
This is a very long standing situation that applies occasionally to only some of the Node-RED users. There are many posts on this subject.
Node-RED comes in two parts. The back end runs on the machine and executes the flows. This will only stop working if it runs out of memory, mainly due to memory leaks or infinite loops. The editor (the bit you are having issues with) runs on your browser and connects back to the back end using a WebSocket.
The editor connects to a server inside the main Node-RED program, and this connection is used to load the flows into the editor, deploy changes, to provide node status and debug information, and to trigger flows using the inject button.
The editor is more likely to fail than the back end. If the editor stops working, this is probably due to issues with the WebSocket connection, which can be down to a number of causes.
Reasons for the editor to stop working (correctly / at all):
more than one Home Assistant (WebSocket nodes) server (overloads the WebSocket)
nodes in the palette set that require updating (corrupt the editor)
servers (MQTT etc) that are too busy for one reason or another
nodes that do not obey the rules (out of date nodes)
flows with existing nodes that need updating to a new schema
issues with the browser / cache used
issues with Home Assistant ingress and/or the iframe environment breaking the connection between the in-frame page and NR across the supervisor proxy
I don’t think that anyone has yet pulled together a complete list of all the ‘solutions’ that have evolved from attempting to fix the issue. I have not had this issue myself, but then I only run a simple set of flows based around the WebSocket nodes, so perhaps turning off all flows (disable all configuration-server nodes) to see if the problem goes away, then turning things back on one at a time to find out the culprit.
Long term solutions do include going away from NR as an add-on. I have a Raspberry Pi 5 running headless with Node-RED, which is very easy to use and connects perfectly back to HA. Editing is just a book-mark click away, and full screen too.