CODEWITHSUNDEEP
node.jsdockerredis
wmbtrmb
wmbtrmb

Reputation: 307

Docker container with redis runned as daemon doesn't run

I have simple Dockerfile

FROM base

RUN <code which installs redis>
RUN npm install redis-adapter

EXPOSE 6379

ENTRYPOINT redis-server --daemonize yes && /app/tasks/redis/entrypoint.sh

And in my entrypoint I'm setting up some configurational keys and set some data to redis via node:

#!/bin/sh

redis-cli hset app:cfg env dev
redis-cli hset app:cfg maxconnections 1024

node /app/tasks/redis/init.js

Image builds succesfully, but when I running it - nothing happens. What's the problem? What should I do to run redis in container and make some configuration after? May be the trouble is in that I'm running redis as daemon?

Upvotes: 1

Views: 1152

Answers (1)

wmbtrmb
wmbtrmb

Reputation: 307

Answer from author

TL:DR

There is a pretty similiar question in Stackoverflow which help to fix my problem:

The problem was in that a Docker ENTRYPOINT or CMD should "spawn single process". And I put Redis starting and node init.js execution as different programs into supervisord. Providing supervisord.conf for example: 

[supervisord]
nodaemon=true
loglevel=debug

[program:redis]
priority=1
command=redis-server

[program:configurations]
priority=2
command=/bin/sh -c /app/tasks/redis/entrypoint.sh

Why did I do that?

The main trouble which I have with this issue was with misunderstanding what actually is a Docker container. And what does ENTRYPOINT or CMD in Docker. I thought that I should just "run some server into Docker and expose some port and Docker will do everything with itself", but that is not the way how containers work. There is a difference between containers and VM's. Look this: How is Docker different from a virtual machine?

When thinking about Docker container as a wrap over one single process it seems clear that code written in my Dockerfile will not work in way that I was expected.

If there is a need to run multiple processes in Docker container then you should use something like supervisord or concurrently (if you prefer Node ecosystem).

Upvotes: 1

Related Questions