The upgrade was successfully completed at around 16:27 UTC, but we're still fighting with some performance issues after the upgrade. Our database and the outbound federation container are currently using significantly higher CPU than expected, which is still being investigated to identify the root cause.
Do you think you guys could do a post on the underlying hardware/infrastructure of .world? I would love to see what it takes to run the largest Lemmy instance lmao
in short form, we have HAproxy running on a hetzner cloud VM for Lemmy/Lemmy-UI, another HAproxy VM for the alt UIs, a VM for the alt UIs.
for the main LW box, which runs lemmy containers, lemmy-ui, pictrs and a database each for lemmy and pictrs, we have a physical box with an EPYC 7502P 32 core CPU, 128GB memory and NVMe SSDs.