After upgrade to new develop, federation problems
my side: receiving HTTP 499 status codes in my nginx logs: inbox posts, timeline gets, anything. their side (according to one person): Error while fetching https://shitposter.club/objects/884282fb-3245-4dd9-b751-82890a31bbfe: {:error, :recv_response_timeout}
I was having a lot of database disconnect, so I adjusted my postgres config and increased the pooling:
pool_size: 35,
timeout: 60_000,
queue_target: 1000,
queue_interval: 15000
config :pleroma, :dangerzone, override_repo_pool_size: true
pgtune gave me this config:
max_connections = 100
shared_buffers = 32GB
effective_cache_size = 96GB
maintenance_work_mem = 2GB
checkpoint_completion_target = 0.9
wal_buffers = 16MB
default_statistics_target = 100
random_page_cost = 1.1
effective_io_concurrency = 200
work_mem = 41943kB
min_wal_size = 1GB
max_wal_size = 4GB
max_worker_processes = 32
max_parallel_workers_per_gather = 4
max_parallel_workers = 32
max_parallel_maintenance_workers = 4
I'm currently doing a repack and analyze on the database so i'm hoping it will speed up but something is causing the http timeout above
here's my Finch config:
config :tesla, :adapter, {Tesla.Adapter.Finch, name: MyFinch}
as you can see it's default, not setting any special config. I do get an error on startup about MyFinch, dunno what that's about.
still getting:
Postgrex.Protocol (#PID<0.5089.0>) disconnected: ** (DBConnection.ConnectionError) client #PID<0.26178.0> exited
I did a timed curl request on one of the posts and it's taking about 20 seconds to return. I added logging to nginx and it appears the clients are cutting off around 5 seconds consistently.