I have Postgres w/ timescale running inside a docker container with ~30 connections to it. I've been noticing that idle connections are using a lot of CPU resources. I'm wondering what would cause that to happen, AFAIK Postgres is pretty good at managing idle connections. If anyone has any recommendations on how to optimize this I would be really grateful!

Hardware specs: 8 CPU cores, 16GB RAM

Connections are all made using postgres.js with default settings

postgresql.conf max_connections = 100 shared_buffers = 3997MB work_mem = 5116kB maintenance_work_mem = 1998MB dynamic_shared_memory_type = posix effective_io_concurrency = 256 max_work_processes = 27 max_parallel_workers_per_gather = 4 max_parallel_workers = 8 wal_buffers = 16MB checkpoint_completion_target = 0.9 max_wal_size = 1GB min_wal_size = 512MB random_page_cost = 1.1 effective_cache_size = 11991MB log_timezone = 'UTC' autovacuum_max_workers = 10 autovacuum_naptime = 10 shared_preload_libraries = 'timescaledb,pg_stat_statements' max_locks_per_transaction = 128 timescaledb.max_background_workers = 16

htop | ID | USER | PRI | NI | VIRT | RES | SHR | S | CPU% | MEM% | TIME | Command | |----|-----------|-----|----|-------|------|------|---|------|------|---------|-----------------------------------------------------| | 14 | dockermod | 20 | 0 | 4230M | 130M | 106M | R | 86.8 | 0.8 | 2:57.60 | postgres: server jerviswebb 172.18.0.28(51524) idle | | 24 | dockermod | 20 | 0 | 4228M | 121M | 101M | R | 86.2 | 0.8 | 1:27.86 | postgres: server jerviswebb 172.18.0.28(48178) idle | | 99 | dockermod | 20 | 0 | 4224M | 127M | 103M | R | 74.2 | 0.8 | 0:37.44 | postgres: server jerviswebb 172.18.0.28(59758) idle | | 78 | dockermod | 20 | 0 | 4232M | 125M | 100M | R | 67.3 | 0.8 | 2:05.81 | postgres: server jerviswebb 172.18.0.28(36496) idle | | 72 | dockermod | 20 | 0 | 4224M | 129M | 105M | R | 65.4 | 0.8 | 1:24.65 | postgres: server jerviswebb 172.18.0.28(57110) idle |