flask -python project with deepface/sqlalchemy database with sqllite engine
d5ff4425-758d-493d-8d72-12a89b68dbac
46 Replies
Project ID:
d5ff4425-758d-493d-8d72-12a89b68dbacwhats your start command?
gunicorn main:app
build logs please https://bookmarklets.up.railway.app/log-downloader/
i asked for build logs, close enough lol
my bad
i will send 1 sec
build logs
send your requirements.txt file please
try adding a nixpacks.toml file with this in it to your project
ok it will take like 5 min to deploy , I will keep u updated and thank for the help by the way
sadly it didn't work do u want the build and deploy logs ?
yes please
do u need the link to my repo ?
truthfully, I have never had success with installing libgl in a nixpacks build and have always resorted to using a dockerfile, so please look into building you app with a dockerfile
lmao sent too soon
updated
can u please send me documentation on how to build with docker
you just write a dockerfile and place it at the root of your project and railway will use that, writing a dockerfile is not specific to railway so there's lots of great resources online
I'd normally write one for you but I'm away from the computer right now, so the best I can do is answer your questions
may u take a quick look at my repo , maybe I am doing something wrong , if there is nothing wrong i will go to online recourses to change to dockerfile
GitHub
GitHub - amr9/Flask-server-face-recognition: this project is implem...
this project is implemented with flask serve back end with two api calls that register the person into a database made by sql alchemy with sqllite engine and another api that find the user in the d...
I don't see anything wrong
ok mr Brody🤝 , I will look into dockerfile if I fail, I will return to you 😂
yes please do
the docker file is made but a new problem came i will send the deployment logs now
send the dockerfile too
not the worst dockerfile I've seen, good job
can you show me a screenshot of your service metrics
why is the image so small
still ?
it looks like your app could be trying to use more than 512mb of ram just the graph isn't showing it because it happens so fast, I'd recommend upgrading to the hobby plan
so the problem is not from dependencies (tensorflow rf)?
beacause 1 of the 3 APIs work fine
well to me those are just warnings about not having a GPU available
ok I will try it tomorrow and give u the news thank you again for the help
yeah let me know if you still have issues when on the hobby plan
the model being loaded into memory is definitely the issue here. You’ll have to upgrade for sure
tensor can run on cpu / mem fine just horribly slow
if it doesnt find the cudnn drivers, i'm pretty sure it won't even try to load ur gpu
Right, but that's not the issue here. Loading the model takes more than 512mb of RAM, their service is crashing as soon as the model is loaded
what kinda model are they loading ?
^
dead link
hence why i ask sorry haha
i see
opencv
yeah that's rough
i have upgraded and it fixed the problem
awesome
thank you all again for the help mr.Bordy this is the third time i came for you for help 😂 
keep coming!