Add healthcheck for default localhost 8008 port on /health endpoint.

[maquis196]

• Updated docker readme for healthcheck related options

Pull Request Checklist

• Pull request is based on the develop branch
• Pull request includes a changelog file. The entry should:
  • Be a short description of your change which makes sense to users. "Fixed a bug that prevented receiving messages from other servers." instead of "Moved X method from EventStore to EventWorkerStore.".
  • Use markdown where necessary, mostly for code blocks.
  • End with either a period (.) or an exclamation mark (!).
  • Start with a capital letter.
• Pull request includes a sign off
• Code style is correct (run the linters)

[maquis196]

I was SO close to using a custom python healthcheck script instead that we could then choose to extend and would avoid adding curl to docker and would allow different base images in the future whilst we know python will always be available. So let me know peoples thoughts.

[babolivier]

Thanks for your contribution! This looks good to me, let's merge it once the CI is done :)

[maquis196]

updated the docker-compose test to make it clear you'll need version 2.1 at least, plus a rebase to see if that fixes the failing CI.

[schildbach]

Just for reference, since 1.20.x I'm using this as my health check from within docker-compose:

    healthcheck:
      test: ["CMD-SHELL", "curl --fail --silent http://localhost:8008/_synapse/admin/v1/server_version || exit 1"]

[maquis196]

Just for reference, since 1.20.x I'm using this as my health check from within docker-compose:

    healthcheck:
      test: ["CMD-SHELL", "curl --fail --silent http://localhost:8008/_synapse/admin/v1/server_version || exit 1"]

out of pure interest, why server_version and not health?

[schildbach]

Ah, I didn't know this endpoint exists. Was using server_version for about a year or so, and it worked.

[SuperSandro2000]

This healthcheck should be reverted. It just delays the image startup by approximately one minute of which the container most likely waits 50 seconds for no reason.

[maquis196]

This healthcheck should be reverted. It just delays the image startup by approximately one minute of which the container most likely waits 50 seconds for no reason.

synapse needs to be running for the healthcheck to work, so what would be waiting a minute? seeing "UNHEALTHY" in a docker ps doesn't mean synapse is uncontactable right?

if i start synapse, i can run the above curl manually within seconds suggesting that synapse is up and running, but docker will wait a minute to say its healthcheck has passed

[SuperSandro2000]

synapse needs to be running for the healthcheck to work, so what would be waiting a minute?

Docker tries the healthcheck once and if synapse is not up yet it waits a minute and then tries again which is annoying.

seeing "UNHEALTHY" in a docker ps doesn't mean synapse is uncontactable right?

If you are using a reverse proxy like traefik it does not get added to the connection pool because it is unhealthy.

[maquis196]

If you are using a reverse proxy like traefik it does not get added to the connection pool because it is unhealthy.

Interesting, so the solution here would be to add '--start-period=5' and maybe change interval to 15, since it tries immediately which is about a couple of seconds too quick

[SuperSandro2000]

That still introduces some seconds delay where you just wait until its done. In my opinion the better solution would be to just remove it as it serves little to no purpose in a single container setup.

[maquis196]

If the powers that be want to remove it, then great, as long as they leave curl in, bottom line is though you can easily disable it on your own setup if you don't feel the need to have it, instructions are in the readme, ill update the new PR with docker-compose instructions on disabling too since i forgot those. (and im not saying you dont know how to, just a good time to offer an alternative)