Post

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

cpu unbothered, moisturized, happy. 8G mem fine (you might not get away with less). of the 72G included disk: - 37G event replay data (6h TTL) - 1.1G postgres - 28G available might bump TTL to 8h but not higher on this disk.

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

ingress? from my other relays: - daily average around 2MB/s - daily typical peaks over 2.5MB/s - bad bunny halftime approached 6MB/s (byte rates! network capacity numbers are usually bit rates) each relay consumer adds a multiple of these numbers for egress, (don't forget overhead for replay)

0 reposts2 likes

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

...if you meant disk usage for client event replay, that 37G implies about 6G/hr to store the raw events. that 6h wasn't over the daily peak though, so it's probably a bit higher