Latest iteration of a page with my home auto, Grafana graph included

Starting to move away from EmonCMS, not because it's bad, but I prefer the Grafana environment where time-shift graphing for comparisons is available...

In that direction, the first "integrated" Grafana with node red:

1 Like

If you are thinking about Grafana embedded into Dashboard, you may want to look instead about pushing data direct into Grafana and cutting out the overheads of Dashboard and iFrames.

You hit the key words...cutting, overheads, direct...

Where can i read up about this?

One of the problems i have is archaic machines....The NR dashboard is about as complex as they can handle, Grafana frames dont display, but the rest is usable...

Granted, optimising wont solve this, but lower overheads will help....

E

There are a few posts in the forum.

Generally though you have a few options.

There are extensions for Grafana that let you post data direct to it via websockets or MQTT so that is one option.

The 2nd option is more complex to set up but easier to manage once done. That is to use InfluxDB and output the metrics you want to it. You can then consume that data directly in Grafana which has built-in support for InfluxDB.

Another option can be useful for showing system stats such as CPU, uptime, etc. That is to use Telegraf along with InfluxDB. Telegraf has lots of options for easily getting data directly into InfluxDB.

Things get a little trickier if you are currently using Dashboard to control things in Node-RED since Grafana is mostly an output tool rather than input. However, there are some extensions that let you create buttons and web links that could trigger flows in Node-RED using http-in nodes.

While InfluxDB and Telegraf obviously add some overheads to your system, there are lots of us that have happily run Node-RED, Mosquitto, InfluxDB, Telegraf and Grafana on old Raspberry Pi's. With InfluxDB, the trick is to make use of its automated capabilities for trimming historical data and for pushing aggregated data to a secondary table - for example, it is common to have detailed (per minute) data for a week but less detailed (hourly aggregated) data for a few years. InfluxDB can do all of that for you with some simple one-off setup commands (the commands are simple, working them out not so much but again, there are posts in the Forum that provide details).

Hi @TotallyInformation ,

Thanks for the pointers!!

I am currently running Influx 1.8 with it being fed directly by Node Red using the InfluxDb Out node, so effectively my setup is pretty much: Node Red with Influx Out for record keeping... There is an EmonCMS node that is running in parallel with Influx at the moment, but that will soon be dropped, once I have my story together properly for Influx and Grafana.

Unfortunately, (Or fortunately), all of my home auto is dashboard driven, with the on-the-spot gauges for what I need to see at a glance, all of the switching done via mosquito and wifi...Yechh... Wifi.... (most of my problems on a day to day are wifi related... But that's another story.)

I have come to realise that the most responsive product for on the spot info is the NR dashboard, the Grafana is a lovely tool for looking back at things, but not so good for live monitoring(But that is probably just my limited experience talking)... And I am happy with that... I only run a single graph on one of all of my dashboards that I use... Even that could be a static jpg of the graph, it shows the day's progress so far, but, alas, I cannot find a way to get Grafana to snapshot it... but, that's another story too...

I haven't looked into Telegraf, would you say that there is a major plus to go and do some research on it? (I am only posting sensor values per se' so nothing really critical in the long term needing speed..Especially since the sensor values are reporting only every 10 sec or so..)

This is pretty much what I am doing...

Unfortunately for me, my archaic pc kit that I use in a few places cannot display the grafana prettiness (XP and the like)...

This interests me quite a bit... It would be pretty handy to have the aggregated data for a decent time span...

Also, from what I have seen of Grafana, its own aggregation methods are pretty damn good! There are often anomalies that show up on them that don't show on Emon....

Wasn't aware of that... Looks interesting, probably more of value towards a "Live Enviro" but worth knowing about..

Enough of my rambling... (Oh look... A squirrel!!)

Cheers
Ed

Know what you mean. I used to have terrible problems in our Victorian house. Until I switched to using better kit. In the form of Ubiquity AP's. Never looked back.

It is actually pretty good though I usually restrict my updates to 1 a minute which is normally plenty. For system monitoring though, I have updates coming in about every 10s:

The trick there as I think I mentioned, is not to keep the detailed data too long.

For me, the major plus is the ability to get system metrics direct into InfluxDB for the dashboards AND, at the same time, dump them to MQTT so that I can use them for flows if I need to. There is so much you can get out of Telegraf with minimal effort. For example, network monitoring as well as system:

Hmm, well, I'd advise changing that old kit over to using Linux instead of Windows, with careful selection of the Linux desktop, you could get quite a performance boost.

This is a 1 yr view from my 5yr hourly data:

Whereas this is a 7d temperature view from my per-minute data:

Or maybe light levels:


Well, between Grafana's charts and InfluxDB's auto-aggregation. When I aggregate from 1 minute to hourly, I keep the max, min and avg values for every measurement.