My debug window only works when i refresh the cache in firefox with crtl+f5... how come?
- what version of NR and node.js
- does it happen in other browsers
- what platform are you running on?
- are you using Docker or Home Assistant?
- when you say "My debug window" do you mean the debug output in the editor sidebar?
- did you make any changes before this started happening?
- Does it happen if you just connect an
injectnode to a debug node?
If you are talking about the debug node, try deleting, adding a new one. deploy, stop and start NR, clear your browsers cache and then see if you get output.
Forgot to mention that if i activate system console in de debug node (not just debug window), i immediately receive the values in the console window of the docker container and not in de browser debug window (side of the flow window) or it takes a very long time to receive them in the browser debug window.
1 what version of NR and node.js : NR version 3.0.2 node.js v16.20.1
2 does it happen in other browsers yes, tested in chrome and firefox
3 what platform are you running on? armbian 23.02.2 with 5.15.93-sunxi64
4 are you using Docker or Home Assistant? Docker
5 when you say "My debug window" do you mean the debug output in the editor sidebar? yes
6 did you make any changes before this started happening? it has been a while sinds i noteced.. but i looked away of the problem... so not sure when this started.
7 Does it happen if you just connect an inject node to a debug node? yes i connect an inject timestamp node to debug node with all "to" options selected.... the only thing that works imediatly is the system console others take very long or don't work.
I just timed the "troughput" of the timestamp to the debug node. the timestamp injects 1691158240671 in the debug node... the console window gives displays this value at 16:10:40 the same timestamp value message appears in the browser debug window at 16:13:16
What hardware is Node-red running on?
What hardware is the browser running on?
the docker is running on a pine 64-LTS
the browser is running
|Processor|13th Gen Intel(R) Core(TM) i7-13700K 3.40 GHz|
|Installed RAM|32,0 GB (31,8 GB usable)|
|System type|64-bit operating system, x64-based processor|
W11
Thanks.
It seems to me like a browser issue rather than Node-red, but since I don't use Docker I'm unsure if that makes a difference.
Can you check the memory usage for Firefox on Windows?
Does it make a difference if you run Firefox in safe mode (Menu | Help | Restart in Safe Mode (or Troubleshoot mode)?
Also check the processor utilisation on the server machine and on the Windows machine.
Memory usage is about 2,5GB but i have a lot of windows open...
did not check iin safe mode but the same thing happends in chrome only less late... injected time stamp at 1691161050372 = 16:57:30 container console window shows the same time stamp for the received message. Chrome time stamp says 16:57:52.
system load of the server is about 33% the windows machine 2%
Since you guys seem to think this is "load" related... im not shure this is related then: when i restart the node red container i always get this message "Started flows
(node:7) TimeoutOverflowWarning: 31536000000 does not fit into a 32-bit signed integer.
Timeout duration was set to 1." Never was able to find out what node 7 is....
There was a bug, but it was solved before Node-red v1.0 !
The work-around back then was stop node-red, delete ~/.node-red/.sessions.json and restart
Since node red is single threaded it has only one core available. How many cores are there and how much processor is node red (or node) using?
the pine A64 LTS uses a Quad-Core Allwinner A64 @ 1.152 GHz. the load average over 1 5 and 15 minutes is 0.75 0.99 and 1.24 .... looks like i'm buying me some new hardware aint it.... ( i used top command to get this info) on forehand i looked at iostat... but that's since the start of the server. Any good advice on convincing the wife i need new hardware?
So... system maybe overload... but is the console getting higher priority than debug window related?
Outputting to the console is much easier than sending data over websockets to a different machine.
Can i somehow check what flows or functions is using that much cpu power? maybe i can simplify my code.
Individual tabs in the editor can be disabled. By disabling a number of them you can work out where the load is. Double click the tab and the enable/disable button is at the bottom.
Start with functions that have for or while loops. Then flows that loop back on themselves.
But the console window runs on the portainer website?
I didn't realise you meant that console.
To prove whether it is a fundamental problem, rename the flows file to something else. On a conventional docker setup I think this is usually /data/flows.json, but check in the startup log to see. Then restart node-red and it will start with an empty set of flows. Then you can add an inject and a debug and see if it is ok. To get back, replace the flows file. You have got that backed up in case of disasters I hope.
