Database issues when following docker install guide

[darkpixel]

The documentation says to run:

docker-compose -p citadel up -d
docker-compose -p citadel -f docker-compose.yml -f docker-compose.migrate.yml run --rm migrate

I ran those with no apparent issues.

The documentation then says:

If you visit http://localhost:3333, you will get redirected to the installation screen so that you can create the first system user.

When I surf to the page, it brings me to a login screen, not a setup screen to create the first user.
This is printed in the log:

postgres_1  | 2020-07-21 20:34:14.735 UTC [27] ERROR:  relation "system_options" does not exist at character 59
postgres_1  | 2020-07-21 20:34:14.735 UTC [27] STATEMENT:  -- name: GetSystemOptionByKey :one
postgres_1  | 	SELECT key, value FROM system_options WHERE key = $1
postgres_1  | 	
web_1       | time="21-07-2020 20:34:14" level=error msg="get system option" error="pq: relation \"system_options\" does not exist"
postgres_1  | 2020-07-21 20:34:14.804 UTC [27] ERROR:  relation "system_options" does not exist at character 59
postgres_1  | 2020-07-21 20:34:14.804 UTC [27] STATEMENT:  -- name: GetSystemOptionByKey :one
postgres_1  | 	SELECT key, value FROM system_options WHERE key = $1
postgres_1  | 	
web_1       | time="21-07-2020 20:34:14" level=error msg="get system option" error="pq: relation \"system_options\" does not exist"
postgres_1  | 2020-07-21 20:34:28.013 UTC [27] ERROR:  relation "user_account" does not exist at character 174
postgres_1  | 2020-07-21 20:34:28.013 UTC [27] STATEMENT:  -- name: GetUserAccountByUsername :one
postgres_1  | 	SELECT user_id, created_at, email, username, password_hash, profile_bg_color, full_name, initials, profile_avatar_url, role_code FROM user_account WHERE username = $1
postgres_1  | 	

Trying to sign in with admin/admin produces this:

web_1       | time="21-07-2020 20:34:28" level=warning msg="user account not found" username=admin

I've tried stopping the docker containers, removing them, and removing the volumes and then running it again with no luck.

[tionis]

I can reproduce this as of 70802ff.

[JordanKnott]

It sounds like the migration step didn't get run successfuly- can you re-run docker-compose -p citadel -f docker-compose.yml -f docker-compose.migrate.yml run --rm migrate and send me the log output from that command?

(after running the above command, try revisiting http://localhost:3333/ to see if the setup screen get's shown)

[darkpixel]

$ docker-compose -p citadel -f docker-compose.yml -f docker-compose.migrate.yml run --rm migrate
Starting citadel_postgres_1 ... done
Error: no change
Usage:
  citadel migrate [flags]

Flags:
  -h, --help   help for migrate

$ 

[darkpixel]

Strange. Even though it says "no change", if I stop and re-start the docker-compose -p citadel up command, it now works correctly.

EDIT: And I had run the migration command multiple times before. I think it maybe just needed docker-compose to go down and up again?

[JordanKnott]

When you re-did the docker-compose -p citadel up command, did it rebuild the image or did it just use an existing one (if there was only a couple lines of output, it used an existing image)?

That's very odd. The no change means that the migrations were run.

[darkpixel]

I hadn't used citadel before.
So I ran the docker-compose up command and it downloaded and built the image. After it was up, I ran the migration and didn't receive any errors. When I surfed to the site, it showed a login page instead of the setup page, so I ran migrations again. It said there weren't any migrations.

I tried blowing away the docker images and the associated volumes and doing it again. That's when I turned in the ticket. But this time, after it said there were no migrations, I stopped docker-compose and started it again and it worked.

I'll have to see if I can duplicate it after removing the images and volumes again.

[JordanKnott]

Well that's just strange - let me know if you're able to duplicate it and the steps you did!

For now I'm going to go ahead and close this as I wasn't able to get it to do the same on my end.

[JadeVane]

Same here. I got a login page when visited http://localhost:3333 after running taskcafe for the first time, so I am not able to create the first user now.

BTW, I build taskcate from source code, and run the command below to try to get it work

# login to postgresql
sudo -u postgres psql

# create database
CREATE DATABASE taskcafe OWNER taskcafe;
GRANT ALL PRIVILEGES ON DATABASE taskcafe TO taskcafe;

# fix a bug
sudo -u postgres psql taskcafe -c 'CREATE EXTENSION "uuid-ossp";'

# run taskcafe
./taskcafe migrate --config config.toml 
./taskcafe web --config config.toml

the bug above means I will get the error message below when I run ./taskcafe migrate --config config.toml follow the documents

Error: migration failed: permission denied to create extension "uuid-ossp" in line 0: CREATE EXTENSION IF NOT EXISTS "uuid-ossp";

CREATE TABLE refresh_token (
  token_id uuid PRIMARY KEY DEFAULT uuid_generate_v4(),
  user_id uuid NOT NULL,
  created_at timestamptz NOT NULL,
  expires_at timestamptz NOT NULL
);
 (details: pq: permission denied to create extension "uuid-ossp")