Post

learned so far: - 4 cores seems _enough to run_ but on this particular host it’s kind of marginal. 16GB ram seems plenty. - it’s been dropping pds websockets, and doesn’t seem to recover on its own. not sure why. - bandwidth is less than i was expecting. - +jetstream on this server is too much

- pds backfill is rough on this server, cpu pegs 100% - made worse by having jetstream on the same server trying to do a higher event rate - also interesting: jetstream just crashes when it’s relay websocket closes? fine if the supervisor auto-restarts it i guess.

mostly unrelated but - i’m still not friends with docker

i think i’m missing some mushrooms. jetstream emission rate is ~330/s where my prod consumer is ~520 where is the best place to get the mushroom PDS list? i went off of this page.

github.com/mary-ext/atp... and I maintain atproto.edavis.dev/bsky-pds.txt from that repo

oh sweet thanks!!

funny, my initial run used Mary’s scraped list, didn’t get any bsky PDSs — i figured i had landed on a bsky-excluded list. errors say different but i think i was self rate-limiting?

mary's list should get you bsky's — here's my crawling script if it's any help github.com/appview-wg-b...

thanks! is that the same from your speedrun gist? bsky PDSs give me 400 HostNotFound…. but actually all PDSs seem to after the first 70-odd on last run. I did crank the PDS/day limit on the relay already. 🤷‍♀️

oh whoops yes it is, already forgot my own blog post that's weird though, is it coming from the pdses or your relay?

hmmm "host server unreachable: server is not a PDS: request failed: Get \"https://oyster.us-east.host.bsky.network/xrpc/com.atproto.server.describeServer\": dial tcp: lookup oyster.us-east.host.bsky.network on 10.89.0.1:53: read udp 10.89.0.12:34729->10.89.0.1:53: i/o timeout"

possibly? i/o timeout is odd — did you try just crawling those and no others

yeah it’s fine until it’s not. they work when i add them manually (like i did with my previous incomplete list) odd indeed. 10.89.0.0/24 is podman’s net, so i guess some kind of podman dns thing in the failing path that i don’t really want to debug. just gonna try slower.

ohh wait no the script fails on all bsky nodes for me. manually adding through web ui worked. whee