We’re seeing cases where one or more assets stall for over 2 minutes, making the platform temporarily unusable.
This behavior appears to have changed within the last two months. It has now become a consistent issue where the platform cannot be used for more than 2 minutes until all assets slowly load and get cached locally (disk cache).
As shown in the screenshot, the assets themselves load quickly when opened directly. However, some asset requests seem to hold onto connections or fail to release them, causing the overall loading process to crawl.
Can we investigate whether specific asset endpoints under public.atl-paas.net are failing to properly release network connections (e.g., holding open or not closing TCP sessions), resulting in Chrome exhausting available connection slots and stalling subsequent asset downloads?
