#32 RFE: Switch from fedora-health-check to its successor
Closed: Fixed 3 years ago by frantisekz. Opened 3 years ago by decathorpe.

Hi, author / maintainer of fedora-health-check here.

I've completely rewritten everything as a service instead of a python script, and it now serves data via an HTTP JSON API instead of storing it in a git repository in pagure.

The code for the new service is here:
https://pagure.io/ironthree/repochecker

For now, I'm hosting it on my own server / website subdomain.
The API endpoints follow this pattern:

https://repochecker.decathorpe.com/data/{release}{suffix}

Where release is {rawhide,32,31} (at least right now), and suffix is either empty or -testing.

The schema of the JSON data is mostly compatible with fedora-health-check, with minor changes:

  • affected maintainers no longer listed separately (closure array content is moved to the JSON root)
  • repo key only contains repository name now, instead of repo : arch+arch+arch
  • addition of the repo_arch key (indicating the arch of the repo vs. the arch of the package),
  • deps array was renamed to be more descriptive (broken)

The default configuration for the service lives here:
https://pagure.io/ironthree/repochecker/blob/master/f/repochecker.toml

The default overrides file is here (copied from fedora-health-check, same format):
https://pagure.io/ironthree/repochecker/blob/master/f/overrides.json

To activate a new configuration or overrides, I only need to run git pull on the server, and it will automatically pick up the new values with the next refresh, so that's nice :)


Cool! From what I can tell, it looks OK to us, as the changes affecting our current usecase are mostly cosmetic. I'll leave confirmation to @frantisekz, but I'm fairly certain.

Thanks for the heads-up, we'll make the switch to the new source of truth (tm) shortly, and we'll close the ticket once the switch is done.

That would be great!

As far as I can tell, it should be better :tm: in every way, with faster data refresh, faster response to requests, and it runs on a server, so it doesn't depend on my home PC and my home internet connection ...

Thanks for letting us know about the change @decathorpe .

I've filled PR with changes to use the new data source: https://pagure.io/fedora-qa/oraculum/pull-request/36 . I'll close the issue once it's merged and deployed.

I have one question though, with previous health check, dashboard uses this as a link to user-readable format of the report: https://pagure.io/fedora-health-check/blob/master/f/reports/report-XX.md .
Is there any human readable version with the new script? Or should we keep linking to the files in the old repo?

Thanks!

I've been thinking about also providing "human readable reports" with the new service, but they don't provide any information that's not already in the raw data.

I can try to migrate the old templates into the new code base and also provide endpoints for getting the reports as pretty HTML, if you think that's something that's worth doing.

PS: The old repo is no longer getting updated, so the files there will start being outdated soon.

I can try to migrate the old templates into the new code base and also provide endpoints for getting the reports as pretty HTML, if you think that's something that's worth doing.
PS: The old repo is no longer getting updated, so the files there will start being outdated soon.

Hmm, okay, I don't think that's necessary. We already show everything in the frontend :)

Thanks

Yeah, that's what I was thinking too.

Thanks for working on this!

Login to comment on this ticket.

Metadata
Related Pull Requests
  • #36 Merged 3 years ago