Performance decreased with last update #4427
Description
As the name suggests following the last update with private and co.
I have a collapse in my performance which creates slowdowns in my access and processing. I have a very large database (4TB) with several libraries.
I go from this (before update):
21:54:49 default_worker.1 | Server: ffdfacd59dc0, Context: sidekiq, Role: background, data: {:cpu=>{:one_min=>0.97, :five_min=>1.18, :fifteen_min=>2.08}, :memory=>0.227037184e9, :disk=>{:available=>159455535104, :total=>268435456000, :used=>108979920896}}
21:55:44 rails.1 | Server: ffdfacd59dc0, Context: rails, Role: web, data: {:cpu=>{:one_min=>1.12, :five_min=>1.19, :fifteen_min=>2.03}, :memory=>0.194113536e9, :disk=>{:available=>158601027584, :total=>268435456000, :used=>109834428416}}
21:55:49 performance_worker.1 | Server: ffdfacd59dc0, Context: sidekiq, Role: background, data: {:cpu=>{:one_min=>1.11, :five_min=>1.19, :fifteen_min=>2.03}, :memory=>0.20082688e9, :disk=>{:available=>158601027584, :total=>268435456000, :used=>109834428416}}
(after update) :
09:19:21 rails.1 | Server: f00d282dd3c8, Context: rails, Role: web, data: {:cpu=>{:one_min=>18.91, :five_min=>18.96, :fifteen_min=>18.63}, :memory=>0.201453568e9, :disk=>{:available=>78874460160, :total=>107374182400, :used=>28499722240}}
09:19:28 default_worker.1 | Server: f00d282dd3c8, Context: sidekiq, Role: background, data: {:cpu=>{:one_min=>17.99, :five_min=>18.76, :fifteen_min=>18.57}, :memory=>0.203395072e9, :disk=>{:available=>78874460160, :total=>107374182400, :used=>28499722240}}
09:19:28 performance_worker.1 | Server: f00d282dd3c8, Context: sidekiq, Role: background, data: {:cpu=>{:one_min=>17.99, :five_min=>18.76, :fifteen_min=>18.57}, :memory=>0.199307264e9, :disk=>{:available=>78874460160, :total=>107374182400, :used=>28499722240}}
So obviously I did the update last night, and I didn"t check the impacts right away.
But here in the state it is unusable each action, takes a crazy time to execute.
To give an example, each page change on a library (page of 6 models with image and not preview stl), takes an average of 6 seconds before it was instant.
I"m not in the code, but I assume that it must check all the permissions by models or even by files, and that it multiplies the access times by "x"?
Environement :
Unraid : Manyfold Docker
Redis + Postgres17
browser : Chrome