Getting 104 Connection reset error after 5 minutes

I have a cluster on GKE and one of my deployments run tornado web app to receive http requests. This deployment is exposed by a LoadBalancer. I send a simple http request to the LoadBalancer ip, which must run on server side for ~10 minutes. After exactly 5 minutes, I get:

requests.exceptions.ConnectionError: ('Connection aborted.', ConnectionResetError(104, 'Connection reset by peer'))

I tried:

• Using the communication locally on my computer (both client and server) and haven't got the reset.
• I made kubectl port-forward directly to the deployment (local client -> kubectl port-forward -> deployment -> server) and haven't got the connection reset. So basically I'm pretty sure it's on the loadbalnacer side.
• I made a backend config with this configuration:

apiVersion: cloud.google.com/v1beta1
kind: BackendConfig
metadata:
  name: my-bsc-backendconfig
spec:
  timeoutSec: 3600

and my loadbalancer is configured like this:

apiVersion: v1
kind: Service
metadata:
  annotations:
    cloud.google.com/backend-config: '{"ports": {"5000":"my-bsc-backendconfig"}'
  creationTimestamp: "2020-12-24T10:08:54Z"
  finalizers:
  - service.kubernetes.io/load-balancer-cleanup
  labels:
    name: wesnapp-flask
  name: wesnapp-flask-service
  namespace: default
  resourceVersion: "14652233"
  selfLink: /api/v1/namespaces/default/services/wesnapp-flask-service
  uid: a922e9cb-4702-481f-b1a9-e09df1653ff7
spec:
  clusterIP: 10.64.9.113
  externalTrafficPolicy: Cluster
  ports:
  - nodePort: 31429
    port: 5000
    protocol: TCP
    targetPort: 5000
  selector:
    name: wesnapp-flask
  sessionAffinity: None
  type: LoadBalancer
status:
  loadBalancer:
    ingress:
    - ip: x.x.x.x

Any ideas how to solve this problem? Thanks