← Back to context

Comment by hobofan

1 day ago

So instead of their self-hosted Gitlab instance being hammered, now their self-hosted Radicle instance will be hammered (and if they are lucky some of the other seeders will tank some of the load)?

I'm not sure that this will actually solve the problem. This seems more like a facade for a move they wanted to do anyways.

> This seems more like a facade for a move they wanted to do anyways.

Not even a facade really. They say this further down in the thread:

> Given our previously communicated desire to migrate to #Radicle, this is a good motivating factor for moving in that direction.

The load will be spread across the network, but I guess the main benefit is that everything continues working even though HardenedBSDs official seed is down.

Every user has their own node, and everyone's node talks to several seed nodes. Even if the official HardenedBSD seed is down, there's still going to be another node to sync with.

  • Does that actually work out in practice? Do you/someone here have experience with that in Radicle?

    IPFS in theory has a similar model, but in practice I've mostly found that if the original seeder goes away, at least part of a dataset becomes inaccessible.

    • I'm not doing anything huge, but my local radicle node is connected to ten other nodes at present, one of which is my own hetzner-hosted seed. Even if half of these go down, I still have full access to all the repos I follow.

    • I think the difference between source repos and arbitrary data objects (which are as often as not images or videos) is that people tend to mirror repos locally indefinitely, especially if it's a local cache of something that they're repeatedly using as a dependency for other software that they develop.

      If anything is good for the bittorrent model, it's git/source control. Movies and images get moved to different drives or deleted, movies become far less worthy of keeping after being watched, and images may have never been useful to the person mirroring them anyway; just a favor they were doing for a site they like. Source code sits, and source code continues to be used. If I understand correctly, Radicle works as your local git server, too.

      The question is whether people will dedicate a little bit of bandwidth to seeding, but I don't think it's a serious question. It's a cheap and easy thing to do if you want to help FOSS, and it's obviously a good and a nice thing to do. It's not like you're seeding stuff that you don't know what it is, or why it is useful.

      And, again, they can keep a seed up indefinitely. But they don't need to have either great uptime or great bandwidth.