[ANN] Hackage will be down for migration today (April 8 2025)

Announcements
1

We will be moving hackage-server to a new datacenter today. Expect extended downtime this evening, and thanks for your understanding!

13 Likes
2

should be done! let us know if there are any problems.

4 Likes
3

Huge thank you to you and the rest of the infrastructure folks for handling this!

1 Like
4

Can we read more on the motivation for this?

That’s more bc I’m curious, not bc I think you have to explain it.

5

Part of the move away from our old infrastructure provider. Our previous provider is sunsetting the product we used.

I’ll be posting a “Quarter 1” update that will include more context.

3 Likes
6

the above answer basically tackles it. i’ll note just that this has been a massive endeavour, and it also raises our regular costs, as our old provider was very generous, and while we’ve cobbled together deals from our combination of new hosts, which we’re thankful for, it still, last i checked, will require more in the way of hosting fees.

basically the whole of ghc build infra has been migrated (you may have seen the announcement about gitlab migration), mail was migrated (which caused some hiccups for some extended time – sorry about that!), and we’re just cutting over hoogle now, and what remains i believe is the main website, the wiki, and the discourse.

of how things stand, the thing i remain least happy with is mail, and that’s because we need someone who knows these types of things to roll up their sleeves and help us make the leap from mailman 1 to mailman 2 – a rather fiddly task.

7

Some Hackage links are redirecting to hackage-content.haskell.org, e.g. Control.Monad.Trans.Class.
Is this part of the migration? Is it permanent?

2 Likes
8

I don’t know if this problem is caused by the move, but I’m getting 502 Bad Gateway on many but not all parts of Hackage. For example https://hackage.haskell.org/package/hypertypes or https://hackage.haskell.org/packages/browse are 502 for me.

5 Likes
9

Yes, when we moved servers we didn’t increase the ulimit on the server, which led to an outage. We’re working to get things back in shape now.

6 Likes