Displaying sensor curves on the dashboard

[pierre-gilles]

@Terdious Ok, super interesting, thanks!

As we can see, the slowest calls are as follows:

Summary

getDeviceFeaturesAggregates.gettingRowsFromLiveDb(energie-totale-phase-1-centre-equitherapie): 1.154s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(puissance-phase-1-centre-equitherapie): 1.155s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(energie-phase-1-centre-equitherapie): 1.157s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(puissance-phase-1-centre-equitherapie): 1.162s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(puissance-phase-1-centre-equitherapie): 1.724s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(energie-totale-phase-1-centre-equitherapie): 1.159s
getDeviceFeaturesAggregates.gettingRowsFromLiveDb(energie-phase-1-centre-equitherapie): 1.156s

getDeviceFeaturesAggregates.gettingRowsFromAggregatedDb(energie-totale-phase-1-centre-equitherapie): 1.733s
getDeviceFeaturesAggregates.gettingRowsFromAggregatedDb(energie-totale-phase-2-centre-equitherapie): 1.742s

These are mainly DB calls on « live » data, meaning the data from the last hour.

Out of curiosity, how many states per minute do you record for these sensors?

Also, as there are many requests, the browser queues the requests to avoid DDOSing the server, so it’s more or less normal, no need to worry

Also, are you running this backend on the same SSD disk as your Gladys production installation? It’s possible that the two are competing for disk access, so you might have better performance in production when you only have one instance

@Terdious

If I take the same example as @VonOx, here if I don’t take the name of the features I would end up with 3x3 times the same device name which no longer makes sense. It’s already complicated because we can’t change the display order. Each « Energie » and « Energie Totale » would therefore be indistinguishable:

I understand the idea.. Maybe these features are just not well distinguishable in the UI (same icon, etc.. we could add something to make it more readable?)

For the distinction between « energy consumed » and « total energy consumed », is this something common in devices found in commerce or is it something you do here @Terdious?

Why wouldn’t « total » be calculated by Gladys?

@Terdious

At the time (over a year ago) I submitted the idea of separating the devices (Type Dual or CH4 from Sonoff) counting several switches into several devices.

This is the solution we are heading towards, as mentioned in the other topic

In any case, after conducting this survey, it’s really the device.name that is important and will be highlighted in Gladys, not the deviceFeature.name.

I will change this view as well, we stay on the device.name

[pierre-gilles]

New build available on the chart tag with the latest fixes:

• Display of the device name instead of deviceFeature.name (following the discussions on the other topic)
• Purely code correction related to the feedback from @AlexTrovato

I still have the NaN bug from @lmilcent to investigate, but I can’t reproduce it for the moment.

Do you have any other feedback?

Does the feature make sense for everyone now?

[pierre-gilles]

Hello everyone!

I’ve been working hard on the PR today, and for me, I’ve reached the end of development:

• I’ve fixed the last things to fix
• Removed the logs added during development for debugging in some cases
• Added data to the demo file for the demo site

I’ve done a final build on the :chart tag as usual.

I’m open to any last feedback, and if everyone is good with it, I’ll merge it into production!

Thanks to everyone who has tested so far, I hope the feature will be useful to many people!

[pierre-gilles]

For information, I just merged this development, which will be available in Gladys v4.6.0 (currently being built)

I will close this topic when the release / release post is live!

Thanks to everyone who provided feedback

Of course, if there are any new feedbacks, I am still open to them (in fact, there will certainly be some with the production release)!!

[pap]

I think that’s great!
I just have a question about the following behavior:
I configured 2 temperature charts (the values from the 2 sensors are sent back to me by the same script via mqtt), and I have one decimal displayed in one case, two in the other. Same for the hours: I have hh:mm on the first graph and hh:mm:ss on the second. Did I misconfigure something?

Moreover, if we take the first graph where the temperatures are displayed with one decimal, shouldn’t the most recent value displayed in bold follow the same format?

PPS: It should be « LastS 24h »
Similarly, if I try to display a duration of 7 days, I get the message « No values in this interval » (intervalLE would be better ); it’s a shame that we can’t display the graph at all on the pretext that we don’t have data for the entire period, I find.

[pierre-gilles]

@pap

I have a decimal that displays in one case, two in the other. Same for the hours: I have hh:mm on the first graph and hh:mm:ss on the second. Did I misconfigure something?

That’s just the library we use that adapts the display based on the amount of data, I’m not surprised at all though

@pap

PPS: It should be “LastS 24h”

Ah! Thanks for the feedback!

@pap

« No values on this interval » (intervalLE would be better )

Thanks too!

@pap

It’s a shame that we can’t display the graph at all on the pretext that we don’t have data for the entire period, I think.

It’s not the entire period, basically when you display intervals greater than 24h, we use the « aggregated » data, because the live data is too heavy to be displayed in such a small graph. The aggregation is calculated only every hour, so it should take a little hour for you to have data

Perhaps a clearer message that specifies this point would be interesting, I think it’s not clear currently

[pap]

ok, impeccable, I’ll wait until tomorrow

@pierre-gilles

It’s just the library we use that adapts the display based on the amount of data, it doesn’t bother me in any case

Indeed, I hadn’t noticed the difference in scale between the two graphs, thanks for the feedback.

[Albenss]

Hello,
Small question about aggregated data, I made a Node-RED flow that retrieves Coinbase data for Shiba. The values are around 0.0005 currently, so the chart on an hour displays well (except for the variation which does not display enough decimal places in my opinion ) but on the 24h, what seems to be aggregated always appears at 0 (I have however apparently activated the keep_history).
Is this normal?

[pierre-gilles]

@Albenss

However, on the 24h view, what seems to be aggregated always appears at 0 (I have, however, normally activated the keep_history).
Is this normal?

You need to be patient, aggregated data is calculated every hour It will come!

I need to clarify this in the UI, you are not the first to ask about this.

@Albenss

(except for the variation which does not display enough decimal places to my taste )

This view was rather designed for home automation display, not so much for such small stock values

[Albenss]

@pierre-gilles

You have to be patient, aggregated data is calculated every hour it will come!

Exactly, it’s been several days (about ten) since my nodered flow has been connected to Gladys, and still no data in « Last 24h »…

@pierre-gilles

This view was rather designed for home automation display, not for stock values as small as this

Ah ah yes I can understand

[pierre-gilles]

@Albenss

Justement, ça fait plusieurs jours (une dizaine) que mon flow nodered est en lien avec Gladys, et toujours pas de data dans “Dernière 24h”…

Ah, I think I understand now! There are indeed values, just that they are at 0 (you can see the line at the bottom of your graph), because for the aggregated values we take an average… I admit that I did not design this screen at all for stock values ^^

I don’t know what the solution would be, maybe suggest different aggregates other than the average?

[pierre-gilles]

The feature has been available since version 4.7, so I’m closing this topic to free up credits!

Nouvelle version 4.7 : Affichage graphique et mise à jour majeure de l'intégration Zigbee2mqtt!

Salut à tous ! Après des mois de travail, je suis heureux de lancer aujourd’hui Gladys Assistant v4.7 (techniquement, Gladys Assistant v4.7.1), une mise à jour assez importante avec 2 principales nouveautés : La vue “graphique” sur le tableau de bord Une mise à jour majeure de Zigbee2mqtt, grâce au super travail de @AlexTrovato. Pour en savoir plus, j’ai détaillé toutes les nouveautés dans un article sur le blog: La documentation a été mise à jour aussi: Afficher un graphique sur le table…