westandskif/rate-mirrors

Manjaro mirror data file change (url and content)

Opened this issue · 7 comments

fhdk commented

The mirror manager project has been in public testing for a couple of months - I don't know the actual change into production - but a heads up is always nice

At some point in the near future expect Manjaro mirror data to change.

The format has changed slightly as can be deducted from https://mirror-manager.manjaro.org/status.json

westandskif commented

Thank you for the heads-up! Just to confirm that I understood correctly, once the thing goes live, I need to switch from:
from https://repo.manjaro.org/status.json
to https://mirror-manager.manjaro.org/status.json
don't I?

And another question is: are you anyhow related to the release of the thing? It would be great if anyone could ping me here when it's time to release the update.

fhdk commented

I need to switch from:
from https://repo.manjaro.org/status.json
to https://mirror-manager.manjaro.org/status.json

It is correct.

And another question is: are you anyhow related to the release of the thing?

I am not involved directly - more indirectly as I supplied some initial testing mirrors in varios stages of sync'ness for the main architect to use. As the principal developer and maintainer I also refactored pacman-mirrors to work with the new data structure.

When it is put to production I don't know - let me ping you when I know it for certain.

  • 👍1
westandskif commented

@fhdk do you have an idea of what timezone "last_sync": "2023-10-17 15:25" is in? UTC?

And just like this issue states, the most recent updates are dated Oct 17th. Either this is going to fixed or we may need to rethink defaults (current rate-mirrors manjaro --max-delay=86400 filters out all the mirrors) or start fetching state files directly on a per-mirror basis.

fhdk commented

For what I could gather from the source, the data is formatted from the mirror's respecitve the branches state files which has a timestamp which - when comparing the timestamp and the file time for the statefile indicates the timestamp is UTC

# Date and time of the last branch change.
date=2023-10-22T00:20:56Z

But the issue is ongoing as it could be a bot configuration error - I have pointed the devs to my ideas - we will see when time allows them to look at it.

  • 👍1
westandskif commented

Thank you, @fhdk ! Either way, the update seems to be ready, let it just sit there until it's needed - #52

westandskif commented

update: just checked https://mirror-manager.manjaro.org/status.json, still reports outdated last_sync (all in Dec 2023).

fhdk commented

Yes - I think the maintainer has been busy with other tasks.

Unfortunatly I have not been able gain knnowledge on when it will go into production 🤷

At the time I created this issue - it was burning hot - now - not so much - I think there has been some issues with the backend daemon failing from time to time - but I don't know much more.

  • 👍1