If I post a new video, or update an existing one, or link to a video the FediDoS can take us offline for 5+ minutes. Usually it's just a blip, but sometimes it crashes a process and we have to start things back up.
I'm sure there are things I could do to make it less extreme, but I haven't tried yet.
@ajroach42@angiebaby@newellijaytv This should scale with "number of instances occupied by followers" as I assume (hope?!) that the previews are generated per-instance.
Unfortunately the post doesn't share the number of servers hitting them (surely it's not 37000+ instances?)
Sites that can't handle this number of requests could presumably take some steps themselves (e.g. "slow down requests from Mastodon instances"). What do they expect "mastodon" to do here, exactly?
@benchaotica@ajroach42@angiebaby There's currently ~11000 mastodon instances; federation itself takes time, so even if you had 11k followers each on their own instance ... is that really "DoS" territory?
On the flip side, do sites that can't handle (at most) 11k requests over "many minutes" already have a way to opt out of preview generation?
Also, "it's foss" takes 84 requests and 6.26MB (or "4 Dooms") to load, perhaps something can be done there too...
@meejah@benchaotica@angiebaby the problem is significantly worse than one request per instance. The details are covered in one of the linked github threads. I'm not interested in continuing this conversation.
Add comment