portainer: Regression with TCP Endpoints in 1.17.0

With 1.16.5, everything works fine with endpoints defined by IP address: image With 1.17.0, something is assuming this is a TLS connection, and failing with the error:

Failure
Unable to connect to the Docker environment

Logs:

[root@vswarm-2 ~]# docker logs -f c37d71dd0bff
2018/05/13 04:05:32 Instance already has defined endpoints. Skipping the endpoint defined via CLI.
2018/05/13 04:05:32 Starting Portainer 1.17.0 on :9000
2018/05/13 04:06:08 http: proxy error: tls: oversized record received with length 20527
[root@vswarm-2 ~]#

Port 2375 is not a TLS port, and TLS is not checked in the settings above. The stack config file is as follows:

version: '3.2'

services:
  portainer:
    image: portainer/portainer:1.16.5
    deploy:
      placement:
        constraints:
          - "node.role == manager"
    volumes:
      - type: volume
        source: portainer
        target: /data
        volume:
          nocopy: true
    ports:
      - "9000:9000"
    command: [ "-H", "tcp://10.61.0.201:2375" ]

volumes:
  portainer:
    driver_opts:
      type: "nfs"
      o: "addr=10.254.0.198,nolock,soft,rw"
      device: ":/ssdpool/docker/voip/portainer"

Docker Version: Docker version 18.03.0-ce, build 0520e24 running on RHEL 7.4

Docker is started with a systemd override to expose the API

[root@vswarm-1 ~]# cat /etc/systemd/system/docker.service.d/override.conf
[Service]
ExecStart=
ExecStart=/usr/bin/dockerd -H 0.0.0.0:2375 -H unix:///var/run/docker.sock

[root@vswarm-1 ~]#

About this issue

  • Original URL
  • State: closed
  • Created 6 years ago
  • Reactions: 1
  • Comments: 41 (17 by maintainers)

Commits related to this issue

Most upvoted comments

Was facing a similar issue to @xrobau. I have tried portainer/portainer:pr1909 and I can now update the stack once more.

Hi @deviantony. I had the same issue after upgrading to 1.17. Since I don’t have that many endpoints I recreated them and after that they’re working fine. After the upgrade portainer wanted to use TLS I got the same error as @xrobau though endpoints were configured to use TCP. After deleting and recreating the endpoints they’re working fine again.