Docker Healthcheck

[markusa380]

Is your feature request related to a problem? Please describe.
It's harder than necessary to detect when OpenSearch Docker containers are ready to accept requests, for example when using testcontainers.

Describe the solution you'd like
I'd like to see the official Docker images use the HEALTHCHECK instruction (which should probably request /_cat/health?v=status) so other systems can depend on it.

Describe alternatives you've considered
Alternative solutions revolve around the systems using the image, e.g. by implementing custom health check loops or, like testcontainers with ElasticSearch, by using a regex on STDOUT.

Additional context
n/a

[reta]

@markusa380 please check https://github.com/opensearch-project/opensearch-testcontainers/

[gaiksaya]

[Triage] We will look into this. Looks like a good thing to have. Thanks for creating the issue @markusa380

[bbarani]

@markusa380 Thanks for opening the issue. We are looking in to this issue. In the meantime, feel free to contribute this enhancement.

[Tokesh]

Hello! I'm newbie at Open Source Community. If it still relevant issue, can you provide information about the dockerfiles in which the heartbeat status needs to be added?
I can try to solve this issue. I'll do my best!

[dblock]

Is opensearch-project/opensearch-go#315 relevant?

[bbarani]

@prudhvigodithi @peterzhuamazon @rishabh6788 @gaiksaya Do you have some inputs for @Tokesh ?

[gaiksaya]

Hi @Tokesh

These are the relevant docker files which would need health check:
https://github.com/opensearch-project/opensearch-build/tree/main/docker/release/dockerfiles

Associated docker images:
https://hub.docker.com/r/opensearchproject/opensearch
https://hub.docker.com/r/opensearchproject/opensearch-dashboards

[prudhvigodithi]

Hey @Tokesh welcome to the OpenSearch community. Here is some useful link, we can add a health check with OpenSearch API that shows the health of the container. For helm charts this is how its done you can use this as reference.

[Tokesh]

Thank you all for the details! I am currently taking my college exams. I will try to send the PR for consideration within a week. I don't seem to be blocking anyone.

[prudhvigodithi]

All the best for your exam @Tokesh.

[peterzhuamazon]

Thank you all for the details! I am currently taking my college exams. I will try to send the PR for consideration within a week. I don't seem to be blocking anyone.

Sure @Tokesh , all the best to your exam and feel free to ask any questions regarding this issue whenever you need.
Thanks.

[Tokesh]

Now I'm thinking that maybe I made the wrong approach for implementation in my PR. I just added a health check to each service in the docker compose.
Perhaps I need to write an additional service in the docker compose that will check the health of other containers through ping/curl at runtime?

[prudhvigodithi]

The best practice to add health check in docker is to use HEALTHCHECK in dockerfile.

Example (using https://opensearch.org/docs/2.7/api-reference/cat/cat-health/)

HEALTHCHECK --interval=30s --timeout=3s \
  CMD curl -f http://localhost:9201/_cat/health?v || exit 1

The above once the image is built docker ps should return the status of the container based on the output of http://localhost:9201/_cat/health?

Also FYI this will not impact when running docker in Kubernetes as Kubernetes doesn't use Docker Healthchecks. They are explicitly disabled within Kubernetes.

@Tokesh can you please explore HEALTHCHECK in dockerfile?

Thank you

[Tokesh]

The best practice to add health check in docker is to use HEALTHCHECK in dockerfile.

Also FYI this will not impact when running docker in Kubernetes as Kubernetes doesn't use Docker Healthchecks. They are explicitly disabled within Kubernetes.

Okkie, i got it! I am very glad that I can learn the best practices. Thank you very much! Made a PR to check. Im ready to change or add if that necessary.

[bbarani]

@Tokesh Any update on this issue?

[Tokesh]

@Tokesh Any update on this issue?

Hello! As I remember, we got a little stuck on the implementation itself. The question was that we need authentication.
If you help me, I would be very happy. Just recently came back home after small vacations. If you give a little more context, what should I do, I can do.

[bbarani]

@prudhvigodithi @peterzhuamazon Can you help @Tokesh here?