I can't say what is what. I am just saying what I see.
I was doing something on that NUC and for the sake of wanting to check, I loaded my browser and looked at NR running on it.
NR has/had ui_LEVEL installed. Not used, but installed.
Other machines have/had that and suffered a similar fate.
For the sake of sanity checking, I removed the node - completely - and rebooted.
Reloaded the browser and looked at NR.
CPU load went from 98% to 30-40%.
Though I usually access their NR via the web from this machine. (Nuc#1)
So though sometimes I run VNC, I try not to and access as much as possible via NR.
At a pinch I use SSH and if I really must, VNC.
But they are RPIs. They don't really like getting that kind of load.
This example is NUC #2. And it is not a toy for the CPU and memory.
Okay, back to the start... you mention "Raspberry Pi locking up after updating NR" in your title and start of the first post. The last messages are about what you call a NUC (I've never heard of that before so I don't know what kind of machine it is) and the screenshot you posted shows username dog on hostname nas-t-nuc so I guess that's the machine you're talking about now. I see a load average of 5.13 1.92 0.72, which suggests that in the last minute it started to do something incredibly heavy compared to before, but without any specifications about the specifications of the machine I've not a clue if that's a bad thing. The machine is indeed running a full desktop environment with Gnome, and based on the colouring scheme I'm wagering a guess that it's running Ubuntu.
A second screenshot is added 2 minutes later, showing a load average that is dropping after deinstalling ui_level. But this is no raspberry pi, by far not looking at the amount of memory available. Set to 8gb of RAM total?
As you mentioned this topic is taking a turn from the Raspberry Pis to those NUCs, but without having any information about these machines we can't really help you. Start with a few simple things:
What OS do they run, is it indeed Ubuntu and what version?
Couple specifications about the hardware, to be able to exclude things related to that: what CPU is inside, is it indeed 8GB of memory as it appears from the screenshot?
Versions/info of a few packages, run the following commands in your terminal. node --version, npm --version, which node, which node-red, node-red --version
Just for confirmation, you have several raspberry pis, throughout the last month I've seen quite a few in hostnames alone. I've seen MusicPi, BedPi, the telemetry thing that I'm not sure where it runs... Now these two NUCs. Can you confirm, for clarity only, that the NUC #1 is what you use as "regular system", for example when accessing NR running on the Pis, or SSH'ing into other machines. And that NUC #2 is a different machine, and the screenshots above with hostname nas-t-nuc are from NUC #2? So for the list of questions above, assuming that the screenshots are indeed from NUC #2, I would like those answered specifically for NUC #2.
The original problem was observed on RPI. It was abut the time I was getting all the machines brought up to date with things. O/S, Nodes, etc.
Some were fine a few didn't want to play the game.
And some just spat the dummy and NR won't run any more.
While doing this I noticed on the ones which were hemeraging (?spelling?) it was . . . strange.
One in particular also had weird dashboard behaviour.
That's w e i r d behaviour.
Screens weren't being shown, and . . . . . other stuff.
I also noticed an unknown "field" of text on the screen.
I traced it down to ui_LEVEL.
I deleted it and uninstalled it.
Rebooted and the CPU load went from 98% to about 40%.
Huh?
I reported it.
By this stage I had gone through all the other machine and removed the node.
I don't need possible problems when trying to get 2 (or 3) machines back to working.
As luck would have it I needed to do something with the NUC (small square/cube shaped computer which run off a plug pack. Similar to a RPI but with a lot more oomph)
It was not playing nice and so I looked at NR on it, (From it) and saw I also had ui_LEVEL installed there.
I uninstalled it and voila! Load went back to about 40% (from 99-104%)
The fact that this machine is head and shoulders above a RPI and simply removing ui_LEVEL is really weird.
I posted screen shots.
Granted I may have highlighted the wrong task. ($hit happens)
When I do top on the RPI, node-red is way up there 98% CPU load.
Dunno why it didn't happen on the NUC.
I am simply reporting what I see.
Just now it is turned off. To get the specs will mean unplugging the keyboard, monitor (etc) powering it up to say it is a NUCm intel i5 from memory (Celeron) with a LOT of memory running ubuntu 18.04. Errrr, what else do you really need?
It isn't a RPI. It is way more powerful. When I turned it on CPU load was unexpectedly high. Uninstalling ui_LEVEL CPU load went down.
This kind of confirms that (despite the machine) something is going on with that node.
Total ram: LOTS. Gigs as you pointed out.
So, yes: The original problem is that on my RPI(s) they were being killed by someting.
The NUC confirms suspicions.
Just now I am overwhelmed with problems. I shall try to remember next time I have said NUC powered up to get information about it.
And all the extra you request.
(space left for separation)
( )
Yes, I have several RPIs.
The ones with problems are:
MusicPi
PiFace
BeefPi
TimPi, TelePi and BedPI are working as far as I know quite well.
THIS machine I am now using is a NUC (#1)
Intel I5 7260 2.2 Gig 4 core. (ok, so the other one isn't an I5. Probably a Celeeron)
16 Gig of memory.
Ubuntu 18.04.3 LTS
For each of my devices, every so often I make an image dump of the SD card. Then I keep careful notes of everything I do to the system. For node-red itself I control the setup using git (I don't use projects to do it but that is a good option). Then if something serious goes wrong I can easily get back to a known working node-red setup by rolling back in git, or even by restoring the card, getting the latest node-red setup from git, and re-applying any system changes I made from my notes.
That way I don't get into the situation you seem to keep finding yourself in where things stop working and you can't easily get back to a working system again.
You keep repeating ui_level, but looking at your screenshots, it does not appear to be ui_level that is causing this cpu behaviour, but I suspect it is your OS. You run a full blown desktop, with a full blown firefox on a raspberry with VNC and NR has to run the flows, with -who knows what- content.
See the "Web Content" utilisation, similar before/after removal.
Before you disconnect from your Pi's make sure to close all the desktop applications.
I added 10 ui-levels, with 2 trigger nodes, 6 values every 5 seconds, no change in CPU, but for the browser itself it is hard to render as there are 100's of individual elements that need to be re-rendered. @hotNipi I'll send you a messsage for a possible improvement.
@Trying_to_learn, if I understand correctly, you run the browser on the Pi using VNC. If you ran the browser on the PC then you would avoid the possible issue of it being the browser clogging up the Pi rather than node-red.
@Colin
Not quite.
The RPI(s) run node red on them.
To access the web page I use FF on this NUC.
The problem I mentioned was all local on a RPI because it was one recently afflicted with problems.
It wasn't practical to do it remotely.
I unplugged the NUC (PC as you call it) and plugged the keyboard/monitor into the RPI to try and first hand sus out the problem.
While doing this, I saw the machine "red lining" for CPU load.
Loading FF and looking at the NR dashboard things looked ok.
But because of what had happened, I went to the edit page and deleted all use of the ui_LEVEL and then removed it.
Reboot may be in there somewhere.
So all that has happened is I removed the node.
Loaded FF and looked. CPU load at a lot lower.
I have nearly got one of them working again so once I get it working and stable, I will re do the test and post what happens.
