Nginx Proxy Manager and Paperless-ngx Issue

[marioteodoru]

I have this Debinan laptop where i run Pi-hole installed baremetal and Nginx Proxy Manager in a docker compose container. It's the only way i was able to do a reverse proxy and also have Pihole running without port conflicts...
I also have Unbound installed and it's configured like this

/etc/unbound/unbound.conf.d/pi-hole.conf

server:
2 # If no logfile is specified, syslog is used
3 # logfile: "/var/log/unbound/unbound.log"
4 verbosity: 0
5
6 interface: 127.0.0.1
7 port: 5335
8 do-ip4: yes
9 do-udp: yes
10 do-tcp: yes
11 do-ip6: yes
private-address: 192.168.0.0/16
62 private-address: 169.254.0.0/16
63 private-address: 172.16.0.0/12
64 private-address: 10.0.0.0/8
65 private-address: fd00::/8
66 private-address: fe80::/10
67
68 # Ensure no reverse queries to non-public IP ranges (RFC6303 4.2)
69 private-address: 192.0.2.0/24
70 private-address: 198.51.100.0/24
71 private-address: 203.0.113.0/24
72 private-address: 255.255.255.255/32
73 private-address: 2001:db8::/32

I made a subdomain at duckdns.org to host my interfaces for Pi-hole and NPM and i can access and log in using a name address like: "https:/www.service.subdomain.duckdns.org" . I think i managed this using letsencrypt and generating SSL certificate, than i added Proxy in NPM and loaded that certificate and than i think i added a local DNS record in Pi-hole for those sites.

Now, i added a bunch of services to the docker-compose.yaml file https://github.com/paperless-ngx/paperless-ngx/blob/main/docker/compose/docker-compose.postgres-tika.yml
And, of course the thing broke.
I can no longer log in to NPM interface or Pihole interface using name address, but i can still log in using IP address with port.
The issue is that now my NPM interface is reseted. It ask me again to log in like the first time and i presume that means it didn't saved the records i made in the docker volumes.

i added
network_mode: bridge extra_hosts: - "host.docker.internal:host-gateway"

to each service in my docker compose stack and now i can see the npm web UI again using name address but it's stuck! It's just the NPM logo with "Loading" underneath. And after a while it says
"The API is not healthy.We'll keep checking and hope to be back soon!"

I guess that this is obviously related to adding new services one of which is db: postgresql.
Is there something related to using db: postgresql in the same stack with npm?
Also this seems to be related to the issue of local IP's.
docker0 has 172.xx.0.1 address. but some services ask for 127.0.0.1 ?