Simplify docker entrypoint of neo4j image

Created by: Tirokk

Authored by roschaefer Aug 25, 2019 Merged Aug 26, 2019

---

🍰 Pullrequest

So we have a permanent downtime of https://nitro-staging.human-connection.org/ and https://nitro-production.human-connection.org/. It's because the database is not running. If you e.g. kubectl -n human-connection exec -it nitro-neo4j-<ID> bash and you run neo4j status then you see: Neo4j is not running.

I couldn't figure out what's the problem. However, when I install htop and run it in the pod, then I see a cypher-shell spinning, it is trying to connect and being terminated all the time:

These two screens ☝ are always alternating. I don't think this is good, because it costs us CPU cycles and money to have a script that's continuously failing and restarting.

The docker image's entrypoint is wrapping the original entrypoint with this script that is trying to setup the indices. So, if the script is restarting in a loop then also this custom entrypoint never exits - even in case of an error. If the entrypoint would fail, then we would also see the neo4j pod being terminated. So that's why I believe it's better to remove the whole thing completely and instead tell our developers to setup the indices manually.

Issues

• None

Todo

• None