Add env vars for seed provider

[hseeberger]

No description provided.

[hseeberger]

The failure on Travis is exactly the same one which I had locally yesterday (#38), but it works for me locally now. Any ideas? Maybe Docker version?

[hseeberger]

Why are the tests still failing? Why is seed_provider null? It's not null on my local machine.

[hseeberger]

Here's the proof:

core@core-01 ~/official-images/test $ ./run.sh hseeberger/cassandra:3.0.0
testing hseeberger/cassandra:3.0.0
    'utc' [1/4]...passed
    'cve-2014--shellshock' [2/4]...passed
    'no-hard-coded-passwords' [3/4]...passed
    'override-cmd' [4/4]...passed

[hseeberger]

Finally CI is passing successfully. @yosifkit, may I ask you to review and merge?

[yosifkit]

I'm a bit on the fence on this. Won't most seed providers require more configuration? It won't be simple to support arbitrary parameters.

This could be accomplished by a user supplying their own config with the provider and its parameters either through -v on run or COPY in a Dockerfile.

[hseeberger]

@yosifkit, I see you point. Could you please explain in more detail how I could set a custom seed_provider?

[yosifkit]

Sure! 😄 Make a local copy of the cassandra.yaml and modify the seed_provider. Then you have two options:

1. if it is local, you can just docker run -v /path/to/cassandra.yaml:/etc/cassandra/cassandra.yaml -d cassandra

2. if you need to deploy it to many machines, then a Dockerfile next to the cassandra.yaml:

   FROM cassandra:3.1
   COPY cassandra.yaml /etc/cassandra/cassandra.yaml

   • then docker build -t custom-cassandra .
   • now docker run -d custom-cassandra (you'll need to tag and push to a docker registry like Docker Hub so that other machines can docker pull it)

[hseeberger]

Ah, now I understand what you mean with local.
I have actually already created a Docker image based on my PR: https://hub.docker.com/r/hseeberger/cassandra. I might change the approach though to modify the official image by applying the necessary changes.

[tianon]

I'm sorry, but honestly if we're going to do anything more with environment variables it would likely be in the opposite direction (which we won't do due to the backwards compatibility implications) -- modifying a YAML file via shell script is very fragile, and setting complex configuration in this way is not great. For custom configuration, the only/best recommendation is to provide an appropriate cassandra.yaml file to your running instance (via a new image with FROM / COPY in a short Dockerfile, via Docker Configs, via bind mount, etc).

Closing in favor of #160 + docker-library/docs#1317.