Traefik v2 no access to service

coollooc · December 21, 2020, 8:55am

I have a docker swarm running with one manager and two work nodes. Now have a very strange situation. Here is my traefik config

version: "3.7"

networks:
  cluster_network:
    driver: overlay
    ipam:
      driver: default
      config:
        - subnet: 192.168.99.0/24
services:
  traefik:
    image: traefik:v2.3.6
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [ node.role == manager ]
      labels:
        - traefik.enable=true
        - traefik.http.routers.traefik_http.rule=Host(`dev.xxx.com`)
        - traefik.http.routers.traefik_http.service=api@internal
        - traefik.http.routers.traefik_http.tls.certresolver=letsencryptresolver
        - traefik.http.routers.traefik_http.tls=true
        - traefik.http.routers.traefik_http.entrypoints=webs
        - traefik.http.routers.traefik_http.middlewares=auth
        - traefik.http.services.admin.loadbalancer.server.port=8080
        - traefik.http.middlewares.auth.basicauth.removeheader=true
        - traefik.http.middlewares.auth.basicauth.users=admin:xxx
        - traefik.docker.network=ks_cluster_network
    command: >
      --providers.docker
      --providers.docker.endpoint=unix:///var/run/docker.sock
      --providers.docker.exposedbydefault=false
      --providers.docker.swarmmode=true
      --providers.docker.network=ks_cluster_network
      --entryPoints.web.address=:80
      --entryPoints.webs.address=:443
      --accesslog
      --log.level=DEBUG
      --api=true
      --tracing=false
      --api.dashboard=true
      --api.insecure=true
      --tracing.serviceName=admin
      --serverstransport.insecureskipverify=true
      --certificatesresolvers.letsencryptresolver.acme.httpchallenge=true
      --certificatesresolvers.letsencryptresolver.acme.httpchallenge.entryPoint=web
      --certificatesresolvers.letsencryptresolver.acme.email=admin@xxx.com
      --certificatesresolvers.letsencryptresolver.acme.storage=/letsencrypt/acme.json
    ports:
      - "80:80"
      - "443:443"
   volumes:
      - /var/run/docker.sock:/var/run/docker.sock:ro
      - acme:/letsencrypt

Then I run traefik as a separate stack with docker stack deploy -c ./traefik.yml ks.

Then I create another stack for services.


networks:
  ks_cluster_network:
    external: true

services:
  whoami:
    image: containous/whoami:latest
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [ node.hostname == node1 ]
      labels:
        - traefik.enable=true
        - traefik.backend=whoami
        - traefik.http.routers.whoami.rule=Host(`whoami.xxx.com`)
        - traefik.http.routers.whoami.entrypoints=webs
        - traefik.http.routers.whoami.tls.certresolver=letsencryptresolver
        - traefik.http.routers.whoami.tls=true
        - traefik.docker.network=ks_cluster_network
        - traefik.http.services.whoami.loadbalancer.server.port=80
    networks:
      - ks_cluster_network
  erp:
    image: monogramm/docker-dolibarr
    deploy:
      mode: replicated
      replicas: 1
      placement:
        constraints: [ node.hostname == node1 ]
      labels:
        - traefik.enable=true
        - traefik.backend=erp
        - traefik.http.routers.erp.rule=Host(`erp.xxx.com`)
        - traefik.http.routers.erp.entrypoints=webs
        - traefik.http.routers.erp.tls.certresolver=letsencryptresolver
        - traefik.http.routers.erp.tls=true
        - traefik.docker.network=ks_cluster_network
        - traefik.http.services.erp.loadbalancer.server.port=80
    networks:
      - ks_cluster_network

whoami service running perfectly ok with "https://whoami.xxx.com". But erp "https://erp.xxx.com" doesn't work.
Using curl from traefik, it shows (192.168.99.13 is the ip of the erp container):

docker exec -it 95bb curl -v http://192.168.99.13:80                                                                        
*   Trying 192.168.99.13:80...
* TCP_NODELAY set
* Connected to 192.168.99.13 (192.168.99.13) port 80 (#0)
> GET / HTTP/1.1
> Host: 192.168.99.13
> User-Agent: curl/7.67.0
> Accept: */*
> 
after a while
> 
* Recv failure: Connection reset by peer
* Closing connection 0
curl: (56) Recv failure: Connection reset by peer

But network is reachable.

docker exec -it 95bb ping 192.168.99.13                                                                       
PING 192.168.99.13:80 (192.168.99.13): 56 data bytes
64 bytes from 192.168.99.13: seq=0 ttl=64 time=70.196 ms
64 bytes from 192.168.99.13: seq=1 ttl=64 time=31.833 ms
64 bytes from 192.168.99.13: seq=2 ttl=64 time=30.375 ms
64 bytes from 192.168.99.13: seq=3 ttl=64 time=69.206 ms

Traefik log related to erp is only getting a close request after i terminate:

"GET / HTTP/2.0" 499 21 "-" "-" 1229 "erp@docker" "http://192.168.99.13:80" 36890ms

But if I run erp on node2, then erp works perfectly fine. In summary,

traefik dashboard shows everything fine;
if erp is deploy in node1 then erp stuck no error message, but can't access
deploy in node2 then works fine.

i did clean out all iptable rules and restart docker, also recreate docker swarm cluster. Still the same situation.

What could be wrong?

Thanks,

diegodevops · January 5, 2021, 12:55pm

Hi,

traefik dashboard shows everything fine

Missing ports 8080:8080 for dashboard?
I would add both networks to traefik service. I like to split a network for traefik itself an other for services.

2.traefik.backend=erp

Dont mix traefik 2.x with 1.7 ! (remove backend)

if erp is deploy in node1 then erp stuck no error message, but can't access

Problem with labels for same node (whoami + erp). Both are using port 80 and you are not defining the services. Try this...

service whoami -->. - "traefik.http.routers.whoami.service=whoami"
service erp --> - "traefik.http.routers.erp.service=erp"

deploy in node2 then works fine.
No conflict because there are in different nodes workers

node 1 -->. whoami
node 2 -->. erp

Best
Diego

coollooc · January 6, 2021, 2:33pm

Hi Diego,

Thanks for you reply. I did change accordingly as following:

labels:
        - "traefik.enable=true"

        - traefik.http.routers.erp-http.rule=Host(`erp.xxx.com`)
        - traefik.http.routers.erp-http.entrypoints=http
        - traefik.http.middlewares.erp-http-redirect.redirectscheme.scheme=https
        - traefik.http.routers.erp-http.middlewares=erp-http-redirect

        - traefik.http.routers.erp.rule=Host(`erp.xxx.com`)
        - traefik.http.routers.erp.entrypoints=webs
        - "traefik.http.routers.erp.tls.certresolver=letsencryptresolver"
        - "traefik.http.routers.erp.tls=true"
        - "traefik.http.services.erp.loadbalancer.server.port=80"
        - "traefik.http.routers.erp.service=erp"
        - "traefik.docker.network=ks_cluster_network"

but still the same. I did nsenter to swarm manager and node1 which is running erp services.

In node1 which is running erp docker service, i can capture traffic via tshark.

Traefik service has ip: 1 92.168.99.16 erp service 192.168.99.108.

From the captured packets, it looks like first connection is fine then destination unreachable.

Any idea?

Thanks,

diegodevops · January 7, 2021, 11:53am

I dont see nothing wrong in your traefik configuration now.

Could you describe you swarm cluster and how are your VM (virtualbox, cloud server, etc) ?

Also could you attach traefik service logs and service erp logs please?

docker service logs traefik_traefik
docker service logs ks_erp

Topic		Replies	Views
Noob connectivity issues when connecting to backend services Traefik v2 docker-swarm	11	751	April 11, 2024
Unable to get Traefik working on docker swarm Traefik v2 docker-swarm	9	3368	July 24, 2024
Traefik filtering portainer as a "disabled container" in docker swarm Traefik v2 docker-swarm	4	1461	April 12, 2023
Traefik 2.0 Docker Swarm Traefik v2 docker-swarm	6	2034	November 4, 2019
Traefik docker swarm deployment: traefik dashboard 404 error Traefik v2 docker , docker-swarm	19	308	April 15, 2024

Traefik v2 no access to service

Related topics