feat: Zinnia Runtime and Peer Checker module

[bajtos]

• feat: install Zinnia runtime

• feat: install mod-peer-checker

• feat: run peer checker module via Zinnia

• feat: switch base Docker image to full node:18

  We cannot use node:18-alpine because Zinnia does not support that yet.

  I tried to use node:18-slim, but Saturn L2 was not able to start in that system.
  The Node.js project recommends to use node:18, so we should be fine. See https://github.com/nodejs/docker-node#image-variants

• test: move mocha config to config file and add --exit

• test: make test/station.js easier to troubleshoot

• test: fix expected Saturn messages

See CheckerNetwork/zinnia#75

TODO

• Rebase on top of latest main
• Unit & integration tests
• Merge metrics from multiple modules
• Docker
  • Add linux-arm64 builds to Zinnia
    feat: build binary for aarch64-unknown-linux-gnu zinnia#172
  • Rework Dockerfile to use Debian as the base image (Zinnia does not support Alpine yet)
  • Verify that Docker images work, especially on arm64
• zinniad: persist job counter across restarts zinnia#148

Out of Scope

• Access logs of individual Zinnia modules. It's tricky! I'll create a follow-up issue for that.

• Documentation for Core developers

  • how to update Zinnia Runtime to a new version
  • how to update a Zinnia Module to a new version
  • how to add a new Zinnia Module to Station Core

[juliangruber]

over to me 😏

[bajtos]

It seems the Ubuntu Docker images are coming in minimal flavour by default, so hopefully there won't be too much bloat introduced by switching from Alpine to Debian 🤞🏻

https://canonical.com/blog/minimal-ubuntu-released

Minimal Ubuntu is a set of Ubuntu images designed for automated deployment at scale and made available across a range of cloud substrates. They use the optimised kernels and optimised boot process on their target compute substrate. These images have a greatly reduced default package set, without many convenience tools for interactive usage. They are much smaller, boot faster, and will require fewer security updates over time since they have fewer packages installed.

https://hub.docker.com/_/ubuntu

Given that it is a minimal install of Ubuntu, this image only includes the C, C.UTF-8, and POSIX locales by default.

[juliangruber]

Any idea what may be wrong and/or how to troubleshoot?

I'll debug it on my Windows machine 👍 Basically for anything Windows specific you can defer that to me, as the team budget bought my Thinkpad.

[juliangruber]

And if you need to sit at the machine, because it's more complicated etc, I'm happy to start the machine and give you remote access

[juliangruber]

.zip extraction previously wasn't required on windows, so it didn't yet have logic for attaching the .exe extension. Should be fixed 👍

[bajtos]

.zip extraction previously wasn't required on windows, so it didn't yet have logic for attaching the .exe extension. Should be fixed 👍

Perfect, thank you for stepping in and fixing the issue!

Windows tests are passing on most versions, but there is a timeout on Node.js 16.

https://github.com/filecoin-station/core/actions/runs/4869576016/jobs/8684249782?pr=92#step:7:72

  1) Activity
       outputs activity:
     Error: Timeout of 15000ms exceeded. For async tests and hooks, ensure "done()" is called; if returning a Promise, ensure it resolves. (D:\a\core\core\test\activity.js)
      at listOnTimeout (node:internal/timers:559:17)
      at processTimers (node:internal/timers:502:7)

  2) Activity
       can be read while station is running:
     Error: Timeout of 15000ms exceeded. For async tests and hooks, ensure "done()" is called; if returning a Promise, ensure it resolves. (D:\a\core\core\test\activity.js)
      at listOnTimeout (node:internal/timers:559:17)
      at processTimers (node:internal/timers:502:7)

I'll re-run the test to see if this is intermittent. It would be nice to make our tests more robust, but I'd rather leave that out of this PR.

[juliangruber]

Agreed, mostly I just re-run these but a better fix would be great

[bajtos]

@juliangruber I added three more commits to make the tests more robust, easier to troubleshoot and less prone to timing errors. PTAL at the changes and also let me know if I should extract those three commits into a standalone PR.

[juliangruber]

One of my commits wasn't signed, which paused auto merging. Sorry about that!

[juliangruber]

https://github.com/filecoin-station/core/releases/tag/v8.1.0