MMM-RAIN-MAP (new: version 2.x)

chassain 0

This post is deleted!

Jalibu

I think about pre-rendering the map animation once as gif image in the backend using phantom instead of live rendering on the frontend. I’ll test that when time allows…

rubenix

@Jalibu sounds nice!

Fozi

@rubenix thank you so much for your thorough and comprehensive analysis and feedback!
I will now disable the above mentioned 5 3rd party modules nd enablle them one by and check if they’d interfere with MMM-RAIN-MAP. I’ll keep you all posted!

rubenix

@Fozi 6 actually. i’ve posted pages & pages indicator @same line :face_with_stuck-out_tongue_winking_eye:

Fozi

@rubenix sorry, you’re right…my mistake. I just restarted MM and got an update notification for MMM-RAIN-MAP. I’ll pull that and test right away with the new version.

rubenix

@Fozi ok thanks, did not check the mirror today and dont know if i can pull the update; let me tuned mate!

Jalibu

I have a alpha version for a backend-generated (pre-rendered) map.
Are there any volunteers who would like to test it? (Please only people who can withstand high frustration :-D )

rubenix

@Jalibu I can test it out on a virtual machine running raspbian for you if that helps :winking_face:

chassain 0

@Jalibu i can again

Fozi

@rubenix I’ve disabled yesterday all 6 modules, but unfortunatelly the situation didn’t improve. The memory consumption was still increasing over the time and within 1-2 hrs. the MM froze. I guess I have to do the brute force method and disable all modules and enable them one by one.

Jalibu

The serverside prerendering works in theory. The recording is just 2-4 Megabytes and the playback on the frontend consumes nearly no additional computing resources.
The little problem is that prerendering is kind of ultra heavy :-D
My Raspi 3 takes more than a minute for that (with only a few additional modules running) and takes >60MB memory for that.
For those of you having problems due to limited resources with live rendering, the prerendering won’t be the answer I guess :-/

Fozi

I FOUND THE “CULPRIT”!!
Yesterday I tinkered around with a module I tried some time ago (MMM-TextClock) and deactivate MMM-google-routes, as I knew the didn’t play well together. In that context I enabled MMM-RAIN-MAP again, just to see if that would have any effect if MMM-google-routes was disabled.
And Bingo!
Since yesterday evening I had no crash or freeze on my MM. Also the memory consumption appears normal and the processor load is between 5%-20% on average (RPi 3b+)

I’ll keep MMM-google-routes disabled for now, as I gives me no real added value at the moment.
And to be clear, it is is a beautiful module and I love it, but I have no current use case for it. I ran it from the very beginning of my MM build without any issues and it is an eye catcher, too. But in favour of MMM-TextClock and MMM-RAIN-MAP we’ll have a pause ;-)

Thanks to all of you and @Jalibu for helping me!

requiemmg

Hi @jalibu,

I really like this module, but I fear that the memory issue still persists.
Just cloned the repo, the module runs fine for some minutes or up to one hour, then the Raspi freezes. I am running on a 1GB 3B+ with 4 pages.
Anything you can do about it?

rubenix

@requiemmg since I updated the pi to a Pi 4 4GB have no issues at all…came from a pi3 1GB too

Jalibu

@requiemmg
yepp, as @rubenix mentioned, 1GB memory is unfortunately too little :-/
On my Raspi 4, 4GB it just works perfectly fine.

I already tried to optimize the rendering to get it working with limited resources, but without success ¯\(ツ)/¯

rubenix

@Jalibu & @requiemmg was tempted to go for the 8GB one, but I can confirm 4Gb is more than enought; go for it and use the pi3 for another project :love-you_gesture_light_skin_tone:

sdetweil

@rubenix u can make the swap space bigger too, helps in some environments

rubenix

@sdetweil i agree with that and sure it helps; but a pi 4 4Gb goes so smoth!!
pi3 1gb its ok for a simple magic mirror; for a complex one with transitions / animations it needs more gauge!

earnestrichards

Great module. It’s running fantastically on my PI4.