Database initialization is not working

[cpang2]

Tried to create database user and database in docker-compose.yml but it is not working. Nothing gets created. Please advise.

influxdb:
image: influxdb:latest
restart: always
container_name: influxdb
ports:
- "8086:8086"
environment:
- INFLUXDB_ADMIN_USER=USERNAME
- INFLUXDB_ADMIN_PASSWORD=PASS
- INFLUXDB_DB=DBTEST
- INFLUXDB_HTTP_ENABLED=true
- INFLUXDB_HTTP_AUTH_ENABLED=true

[dawidchyrzynski]

Any progress on this? I have the same issue when auth is enabled.

[dawidchyrzynski]

I've fixed the issue by escaping "true" value for INFLUXDB_HTTP_AUTH_ENABLED variable. My compose service looks like this now:

influxdb:
  image: influxdb:latest
  restart: always
  container_name: influxdb
  ports:
    - 8086:8086
  environment:
    INFLUXDB_DB: telemetry
    INFLUXDB_ADMIN_USER: influx_admin
    INFLUXDB_ADMIN_PASSWORD: pass
    INFLUXDB_HTTP_AUTH_ENABLED: "true"

@cpang2 can you confirm whether it works for you?

[repomaa]

Having the same issue. Passing true as a string does not fix it.

[IronSinew]

I am also seeing this issue

environment:
     - INFLUXDB_DB=database
     - INFLUXDB_ADMIN_USER=admin
     - INFLUXDB_ADMIN_PASSWORD=admin
     - INFLUXDB_HTTP_AUTH_ENABLED=true

Database is not initialized. Users are not created.

[mmodenesi]

Same issue, using docker-compose version 1.19.0, build 9e633ef

edit:

Sorry, my bad. Using image: "influxdb:latest" fixed it (I was using an old version before). So I can not reproduce this, but maybe you can confirm (after container is created) that your influxdb image (latest) is indeed a recent one.

[emriver]

As mentionned by @PSanetra in another thread, cleaning my influxdb volume fixed it for me (rm -rf /influxDbData)

[scho]

I ran into a similar issue. I've set INFLUXDB_DB, but no database was created.

There was an issue with the special characters not being properly escaped. In my case, the database was called metrics-live and the problem was the hyphen. I had to add quotes to the name, so that it worked ("metrics-live").

[halfbaked]

thanks to the comment above from @emriver . cleaning the volume fixed it for me.

[blockloop]

My problem was having influxd in my command:. The docker CMD executes a script which is a proxy to the influxd command, which executes all of the init stuff first.

OLD:

  influxdb:
    image: docker.io/influxdb:latest
    environment:
      - INFLUXDB_DB=prometheus
    volumes:
      - ./influxdb.conf:/etc/influxdb/influxdb.conf
    command:
      - 'influxd'
      - '-config'
      - '/etc/influxdb/influxdb.conf'

NEW:

  influxdb:
    image: docker.io/influxdb:latest
    environment:
      - INFLUXDB_DB=prometheus
    volumes:
      - ./influxdb.conf:/etc/influxdb/influxdb.conf
    command:
      - '-config'
      - '/etc/influxdb/influxdb.conf'

[lchachurski]

This is what worked for me using docker-compose. Keep in mind it doesn't work well when changing those settings back and forth. Once the database is initialized it will somehow rebel against changes and output weird errors while authenticating. In my case deleting /tmp/influxdb directory helped.

So delete any database dir first and then use the config below:

    influxdb:
        environment:
            INFLUXDB_DB: project_dev
            INFLUXDB_ADMIN_USER: admin
            INFLUXDB_ADMIN_PASSWORD: secret
            INFLUXDB_HTTP_AUTH_ENABLED: "true"
        image: influxdb:latest
        ports:
            - "8086:8086"
            - "8087:8082"
        volumes:
            - /tmp/influxdb:/var/lib/influxdb

[Surfict]

Hello, still have the problem with the current version.

[cooling75]

... Keep in mind it doesn't work well when changing those settings back and forth. Once the database is initialized it will somehow rebel against changes and output weird errors while authenticating. In my case deleting /tmp/influxdb directory helped.

Thank you for this hint, I struggled with this for hours. Deleting the volume and let docker-compose recreate it solved it.

[JoshEisele]

This issue is quite old - is there any hope that it will be fixed at all. Any workaround to set the ADMIN_USER, ADMIN_PASSWORD as well as the option to HTTP_AUTH_ENABLED at initial docker-compose?

[elboletaire]

Checking the init-influxdb.sh file, I see that the only way to initialize users in influx is by setting both INFLUXDB_HTTP_AUTH_ENABLED and INFLUXDB_ADMIN_USER. At the beginning of the init-influxdb.sh there's an initialization of a variable named INIT_USERS, which is then used to check if db initialization should be made.

INIT_USERS=$([ ! -z "$AUTH_ENABLED" ] && [ ! -z "$INFLUXDB_ADMIN_USER" ] && echo 1 || echo)

I have checked it and it properly initializes the db and creates the user if you specify all the required variables.

I've stumbled upon this issue for a different thing though (I'll create a different issue for it right away): if you specify a command it's completely ignored, even though it's specified in the official readme:

[iosonoumberto]

Hello,
same problem using both latest and 1.8 using docker file

env should be correct (this is from my docker-compose.yml file)

    environment:
      INFLUXDB_DB: jlogs
      INFLUXDB_ADMIN_USER: admin
      INFLUXDB_ADMIN_PASSWORD: admin
      INFLUXDB_HTTP_AUTH_ENABLED: "true"

It should work as I have both ADMIN USER and HTTP AUTH enabled
moreover, inside container, i see the env vars

root@335d30cd03d4:/# env | grep INFL
INFLUXDB_DB=jlogs
INFLUXDB_ADMIN_USER=admin
INFLUXDB_HTTP_AUTH_ENABLED=true
INFLUXDB_VERSION=1.8.3
INFLUXDB_ADMIN_PASSWORD=admin

so it has everything to create user and db...in theory

i can see these logs

[httpd] 127.0.0.1 - - [22/Oct/2020:22:37:28 +0000] "GET /ping HTTP/1.1" 204 0 "-" "InfluxDBShell/1.8.3" 2a348f04-14b7-11eb-8001-0242c0a85002 759
ts=2020-10-22T22:37:28.460603Z lvl=info msg="Executing query" log_id=0Q0cCQ90000 service=query query="CREATE USER admin WITH PASSWORD [REDACTED] WITH ALL PRIVILEGES"
[httpd] 127.0.0.1 - - [22/Oct/2020:22:37:28 +0000] "POST /query?chunked=true&db=&epoch=ns&q=CREATE+USER+%22admin%22+WITH+PASSWORD+%5BREDACTED%5D+WITH+ALL+PRIVILEGES HTTP/1.1" 200 57 "-" "InfluxDBShell/1.8.3" 2a34bf6a-14b7-11eb-8002-0242c0a85002 126506

and these logs

[httpd] 127.0.0.1 - admin [22/Oct/2020:22:37:28 +0000] "GET /ping HTTP/1.1" 204 0 "-" "InfluxDBShell/1.8.3" 2a4a8069-14b7-11eb-8003-0242c0a85002 99
ts=2020-10-22T22:37:28.702907Z lvl=info msg="Executing query" log_id=0Q0cCQ90000 service=query query="CREATE DATABASE jlogs"
[httpd] 127.0.0.1 - admin [22/Oct/2020:22:37:28 +0000] "POST /query?chunked=true&db=&epoch=ns&q=CREATE+DATABASE+jlogs HTTP/1.1" 200 57 "-" "InfluxDBShell/1.8.3" 2a4a9d28-14b7-11eb-8004-0242c0a85002 108868

Anyhow, if i check on my local directory used for influx data i do not see the db created

[root@tig fig]# ls ../influxdb_data/data/
_internal

only internal db is created

as a result, when i go to grafana and try to create a data source it fails due to authorization issues (as i'm probably trying to 'connect' to a db that does not exist

Is there a working tag or something i can do to make it work?

[thomascrha]

Ok the issue for me was with the meta_data directory check with the in the if statement before the init stuff occurs.

if ( [ ! -z "$INIT_USERS" ] || [ ! -z "$INFLUXDB_DB" ] || [ "$(ls -A /docker-entrypoint-initdb.d 2> /dev/null)" ] ) && [ ! "$(ls -d "$META_DIR" 2>/dev/null)" ]; then

The last part is checking if the $META_DIR directory exists - if it does it won't continue. This is important as it's what really see's if the init command have been run and if they needs to be run again. My problem is I don't want to use a volume but a bind - and I can't do that with /var/lib/influxdb - so I bind all the sub dirs to get around this i.e. data, wal and meta. When I do this I create the directory - disallowing the init stuff to work. To get around this I simply check if the directory contains any files rather than checking if the directory exists.

What I did was replaced

[ ! "$(ls -d "$META_DIR" 2>/dev/null)" ]

with

[ $(ls "$META_DIR" | wc -l  | grep -w "0") ]

[andcan86]

Please specify the OS on which you run docker and the docker-compose version you are using.

BTW, it might have to do with the OS, on Windows 10 I have to use double quotes for the environement variables, so:

INFLUXDB_DB="db-test"

and it works. hope that helps!

[suadev]

that's worked for me

  influxdb:
    image: influxdb:latest
    environment:
      DOCKER_INFLUXDB_INIT_MODE: setup
      DOCKER_INFLUXDB_INIT_USERNAME: admin
      DOCKER_INFLUXDB_INIT_PASSWORD: password
      DOCKER_INFLUXDB_INIT_ORG: myorg
      DOCKER_INFLUXDB_INIT_BUCKET: krakend
    ports:
      - "8086:8086"

[sailesh-suri]

If you're still using influxdb:1.8, you'll need to set a custom meta DB directory with the INFLUXDB_META_DIR environment variable. The /init-influxdb.sh script that's supposed to create your admin user account does not seem to function if you use the default meta directory. Not sure why it was designed that way. The below configuration is working for me.

  influxdb:
    image: influxdb:1.8
    ports:
      - "8086:8086"
    environment:
      - INFLUXDB_DB=default-db
      - INFLUXDB_HTTP_AUTH_ENABLED=true
      - INFLUXDB_ADMIN_USER=admin
      - INFLUXDB_ADMIN_PASSWORD=password
      - INFLUXDB_META_DIR=/var/lib/influxdb/meta2

[sakeebs]

Do we have any solution for this, as after every restart of the container or docker service or server reboot I have to change the meta DIR variable.
I am upgrading the influx from 1.7 to 1.8.10

• INFLUXDB_META_DIR=/var/lib/influxdb/meta2