Post

fig (aka:[phil])
@bad-example.com

the primary microcosm api gateway is still running on a digitalocean droplet. hopefully will be able to move it next week so i can close my account there. wish i didn’t always have such a backlog.

0 reposts51 likes

fig (aka:[phil])
@bad-example.com

and in case you’re curious: the gateway is the public access to all the services that run from my house. the secondary runs on a lil vultr vps. the pair are dns load balanced. this is nice because the move should be zero downtime.

fig (aka:[phil])
@bad-example.com

the gateways run nginx and reach my home network via tailscale-- cleartext http over wireguard. it has all just worked super reliably. the gateways also - terminate tls - cache responses for a few seconds - rate-limit by client ip - are small enough to fail before a DoS could ruin my home internet

fig (aka:[phil])
@bad-example.com

these days they serve 250–1,800 req/sec, mostly for constellation (slingshot isn’t hosted at home) dns load balancing isn’t perfect, the volume is often unbalanced between the gateways. this unbalance is good actually, since their purpose is redundancy, not capacity. either can still do full load.

johanna
@jola.dev

Sounds like it doesn’t need a lot of resources but a generous bandwidth allowance would be nice? I haven’t run anything in OVH yet, most of my stuff is on Hetzner. How’s the experience/reliability etc?

fig (aka:[phil])
@bad-example.com

yeah pretty minimal requirements even for bandwidth tbh! no firehoses going through it, and api requests/responses are usually small. low latency to the origin (my house) is helpful, i like hetzner but they don’t have a toronto presence.