Migration error on boot upgrading to v1.5.7 about valetudo_vacuum_camera HOT 7 CLOSED

Thanks for the report @Cellivar I will check the migration process and correct it in a couple of hours.
I think also there is some adjustment done already and the svg_export is a future that somehow is still under development so.. it should be quite easy to fix.

from valetudo_vacuum_camera.

@Cellivar I just tested the migration process from 1.4.8 to 1.5.7.4 (that I'm going to release in a couple of minutes), the config entry 2.1 can be now converted without any problem to 2.2.
From what version did you start the conversion? Also, if you like this project please consider to star it to get always updated on new evolutions ;)

from valetudo_vacuum_camera.

@Cellivar the version v1.5.7.4 is now released. There is now the auto trim function that should simply the setup of the camera on this newer versions. Of course the colour settings can be.. to be redo.. (if the camera was deleted).
In general this fix the migration issue you kindly reported, anyhow I'm at your disposal for anything else and await your kind confirmation.

from valetudo_vacuum_camera.

@Cellivar added to the pre-release check list a test to avoid this in the future. Thanks once again for the hint.

from valetudo_vacuum_camera.

Wow thank you for the really fast fix! I just applied v1.5.7.4 and can confirm that the reported issue is fixed, however I encountered a new problem.

I previously was running Home Assistant with a limit of 756MB of RAM. After updating to the latest version there's a ~1,400MB spike of memory use during startup, then the system settles down to around 780MB. This is what the chart looks like.

I have some extra space in my cluster so I can work around this problem by increasing the limit, but this is a much higher resting state than before. I don't have time this weekend to test further by enabling and disabling things to confirm that it's definitely this integration.

If you have any suggestions on ways I can help you understand what my system is doing I'm very happy to collect more logs or otherwise. I'm a software developer, but I don't write Python code, so I'm not familiar with those tools :)

from valetudo_vacuum_camera.

@Cellivar any help here is appreciated. I do not know from what version you updated so I can't make a comparison.
Recently I added a couple of new functions (auto-cropping is one of them) and also I wrote a function to check in what room is the robot (those are the biggest change).
The auto trimming of the image as the robot position check at startup uses a little more memory than before but the memory should free after that (except for the buffered base image that isn't redrawn in case the LiDAR detects "new walls").

On the v1.5.8 I plan to keep the memory constant to the minimum after the camera boot(this is what I'm actually checking with the new introduction of the memory limit)

This checks the size of NP array multiplied the number of steps to complete the image (as I'm not sure if Numpy is actually freeing the memory). In reality I'm deleting all unnecessary arrays on the draw functions) that I had to write by my own as per home assistant OS do not support OpenCV.

Any logs from the Camera and details of system where you run HA would be helpful to warrant that this project goes ahead improving. Would you be so kind to open a discussion so that we can trace this please 😉

from valetudo_vacuum_camera.

Camera Memory usage in GB: 0.63, 8.24% of Total. (I have to Camera's on the PI4 and this need to be so 1.2GB peak max on my test environment).

from valetudo_vacuum_camera.