There is a brand new Windows 10 PC on the other side of the world for which I'm the go-to guy when things go wrong.
It would be nice if the PC was sending me it's temperature, memory usage, etc so I could spot trends before the ship hits the iceberg.
At home I expect to receive data from MQTT into Node-red and then to a database, but I'm wondering what other people would run on the monitored machine to send this data.
I can install Node-red; Bart Butenaer's node-red-contrib-cpu looks promising, but maybe there are simpler command line options that people can recommend?
Well, you can run Telegraf on Windows not just Linux and that might be a good starting point. Windows itself has a number of built in monitors though some of them may require W10 Pro or Enterprise not W10 Home. I'd start with Telegraf to see if that will capture what you want.
Then you need to get that data somewhere useful. Thankfully, Telegraf supports MQTT output.
Secure communications of some description for the remote PC to send you the data on a regular basis
Method to grab the data on the remote PC, bundle it up and send it over the secure channel
Personally i would create a Zerotier (or Cloudflare) SDN between the two nodes to give you the secure comms channel and then i would run Glances as a low impact way to get the info you want -but as Julian noted Telegraf is good also
There will be zerotier between this system and mine.
I'll take a look at Glances too, thanks.
After a couple of days hurling invective at Windows Powershell (trying to install things) I have renewed my oath never to buy another MS product.
I know I've put Node-red on PCs before but I'm blowed if I can make it work now.
Though I have to say that the sheer volume of spelling mistakes on the website doesn't exactly fill me with massive confidence.
The documentation appears to be better than the website itself though. And they get a for having 2FA built in.
The one thing I don't like is that you have to use a certificate that the system generates (which it does via Let's Encrypt if port 443 is available on your firewall and you allow it to be used for this, alternatively it falls back to a self-signed cert), it doesn't look as though you can use your own certificate. That's annoying for those of us already using Let's Encrypt.
BTW, I'd also recommend not using their default Internet facing ports. Certainly not port 80 unless you really have to. Too many bots probe that port. While it might be OK, best to avoid the issue and use a different port. In any case, you might well need that port for something else.
Not directly Node-red related but a progress report...
Telegraf on Windows has gone through my review process.
Does it work out of the box? No
After I install it for the nth time does it work? No
Can I make it work by following / modifying the instructions? No
Can I find an on-line tutorial showing the current installation and execution process (without Influxdb, which in my case I do not have [nor want])? No
Can I find an example of MQTT used as an output not input filter? No
Does the website annoyingly change my browser tab label every 2 seconds? Yes
All I want for christmas is an operating system with a well designed command line interface!
And another progress report - Julian's alternative Node-red installation on Windows.
Does it work out of the box? Seems to.
But slightly alarming warnings
npm WARN deprecated multer@1.4.4: Multer 1.x is affected by CVE-2022-24434. This is fixed in v1.4.4-lts.1 which drops support for versions of Node.js before 6. Please upgrade to at least Node.js 6 and version 1.4.4-lts.1 of Multer. If you need support for older versions of Node.js, we are open to accepting patches that would fix the CVE on the main 1.x release line, whilst maintaining compatibility with Node.js 0.10.
npm WARN deprecated axios@0.27.0: Formdata complete broken, incorrect build size
Since I have node.js 16.6 I guess this can be safely ignored.
Does node-red run successfully? Yes
Can I connect it over zerotier to Mosquitto broker on a Raspberry Pi back home? Yes
It's looking promising, and there are even nodes available which [may eventually] give me info such as cpu temperature. Disk/memory usage should be a doddle, even on Windows?
Now, how to make Node-red start up automatically on Windows...
I thought I would test out their support by posing this question to them. They responded very quickly which was good news. The answer was that you can get the system to create the internal self-signed cert and then change the configuration manually to use your own cert. So also good news.
Definitely something to test out when time permits
You have a couple of options. There is a tool that will create a Windows service out of an executable. Alternatively, you can configure the Windows scheduler yourself to start up the node-red script. In the 2nd case, I tend to use PM2 as a helper but you don't need to and it does take up some resources, it is simply easier to access the logs and automate restart on break.
I wonder why common tools like incinga, cacti, zabbix or nagios have not been mentioned. These are professional agent based monitoring tools (so no out of the box thing), but if it is for the learning it is definitiely worth the effort.
Ok i know its not node-red based, but I just want to give the hint, that the described task (and a lot more) can be solved professionally with the described tools (and similar others of course as well).
They are all fine products in their own sectors but rather overkill and complex for monitoring the relative's PC's
That's why I like Telegraf because it is easy enough to configure, doesn't take a lot of system resource and integrates easily with both InfluxDB and MQTT.
Yeah great as a learning exercise for all of the tools mentioned - not appropriate for a one off on a relatives PC for staying ahead of basic problems (like running out of disk space, memory issues, Processor overload etc)
for having 2FA built in.