Login on fixed users mode doesn't work

[OfirKP]

After several attempts to define fixed users in apiserver.conf (and trying your example), signing in is unsuccessful.

After deleting and installing the new docker image and containers, I followed the instructions in your latest release to add a list of fixed users (and used no special characters).
I restarted the trains-apiserver container. And yet, when I got to the login screen (the new one, and not the "Enter Full Name" screen), signing in with the correct details gives an "Invalid User/Password combination" error.
The response from the auth.login service is the following:

{  
   "meta":{  
      "id":"802d6a6810d24c1a885292ef35c358dc",
      "trx":"802d6a6810d24c1a885292ef35c358dc",
      "endpoint":{  
         "name":"auth.login",
         "requested_version":"2.2",
         "actual_version":"1.0"
      },
      "result_code":401,
      "result_subcode":22,
      "result_msg":"Unauthorized (invalid credentials) (failed to locate provided credentials)",
      "error_stack":null
   },
   "data":{  

   }
}

I get the same error when using your example with username: jane, password: 12345678.

[bmartinn]

Hi,
This looks like a configuration loading issue (type of error indicates the fixed users configuration was not located by the server).
Can you share some environment details:

• apiserver.conf file contents (even if it's the same as described in our example)
• apiserver.conf file location in your file system
• docker run command used to start the trains apiserver docker container

[OfirKP]

Thank you for the response!
I would like to mention that after deleting apiserver.conf and restarting the trains-apiserver container, I'm getting back the regular "Full name" screen and not the username & password login screen.

Here are the details:

File location: /opt/trains/config/apiserver.conf

Content

auth {
    # Fixed users login credetials 
    # No other user will be able to login
    fixed_users {
        enabled: true
        users: [
            {
                username: "jane"
                password: "12345678"
                name: "Jane Doe"
            },
            {
                username: "john"
                password: "12345678"
                name: "John Doe"
            },
        ]
    }
}

Docker command used: sudo docker run -d --restart="always" --name="trains-apiserver" --network="host" -v /opt/trains/logs:/var/log/trains -v /opt/trains/config:/opt/trains/config allegroai/trains:latest apiserver

[bmartinn]

Looks OK. This might be caused by incomplete data left in the database on server initialization... Can you share the server logs (in /opt/trains/logs)?

[OfirKP]

I think you are right. Can you elaborate further on how I could reset/remove that data? (I tried to delete the opt/trains folder but it didn't help)

Logs:
apiserver.log
apiserver(1).log

[bmartinn]

The database data resides in the /opt/trains/data folder. To delete it and start over:

• Stop and remove all docker containers (as described in the Upgrade section, step 1)
• Delete the contents of the /opt/trains/data folder
• Delete the logs from /opt/trains/logs (just to make it easier to see what's going on)
• Run the docker containers

[OfirKP]

It failed after reinstalling the containers (and deleting the folders as you suggested), but I managed to figure out the problem.
After installing trains on a completely new environment I still got the same issue, and when I ran docker ps I found out that the trains-mongo container was stuck in a restart loop, because of a conflict it had with a running instance of mongo on my host with the same port (27017).

Killing the host's mongo solved the problem.

However, I would prefer not to change my host's mongo defaults.
I manually resolved this by adding the command mongod --port 27018 (or other unused port) to your docker run command.
e.g. sudo docker run -d --restart="always" --name="trains-mongo" -v /opt/trains/data/mongo/db:/data/db -v /opt/trains/data/mongo/configdb:/data/configdb --network="host" mongo:3.6.5 mongod --port 27018

Also, I had to change the 27017 port to 27018 in the trains-apiserver's server/config/default/hosts.conf to make it work.
Is there a more convenient way to override this configuration without manually editing this file? (and one that will survive an update)

[bmartinn]

Glad to hear you figured it out, and thanks for the update! - we'll add it to the FAQ.
As for configuring an alternative port number - I'll add a feature request to allow setting this value using an environment variable, which will make it easier to control using the docker run command.