Hey team, we're seeing some interesting behaviour from the tracing of our CheckPermission calls where SpiceDB is doing nothing and hanging, increasing our latency. Attached is a screenshot. To give some context, we're self-hosted, we use the SpiceDB operator and we're using Postgres. Can someone help inform what might be happening here? Thanks a lot in advance! https://cdn.discordapp.com/attachments/844600078948630559/1387149098427093164/Screenshot_2025-06-24_at_1.10.08_PM.png?ex=685c4ac5&is=685af945&hm=ce5a824b9c3ddcfa80e261485f3d5a2aebb9cee6f5aeab80c8ab5cebf8b8e9c1&

the first thing i'd check is whether spicedb is waiting on DB connections

Hmmm but it's not waiting before some operation, it looks like all the subtasks are done and yet it's still not closing? https://cdn.discordapp.com/attachments/1387149098766958664/1387149568982126677/Screenshot_2025-06-24_at_1.16.42_PM.png?ex=685c4b35&is=685af9b5&hm=9e65603c8d2c7583bb8fa785b0332b716dab3b30c4c0d426624218949307f7d3&

pgxpoll_empty_acquire

is the metric that SpiceDB exposes that will tell you whether SpiceDB was waiting on connection acquisition

pgxpoll_empty_acquire is looking good from what I can tell... We've also setup pgbouncer https://cdn.discordapp.com/attachments/1387149098766958664/1387149989905567794/Screenshot_2025-06-24_at_1.18.28_PM.png?ex=685c4b99&is=685afa19&hm=5cfc64c2c4856a7554cda6ba8fe657e66aa09a4aac56fccaec27ee31ed5096ca&

CPU usage being high could be related

🤔 How do you recommend we proceed? 😢

are you familiar with the go profiler? spicedb exposes a [profiling endpoint](https://authzed.com/docs/spicedb/ops/observability#profiling) that could help you understand where it's hanging

Will give that a try, thank you!