Locally cached HTML is grabbing from service worker for 1+ second(s)

Seeing the following picture on a starter project when I do PLP or PDP browsing transition:

Why Resource Scheduling can take so long?
In this case the data requested was 100% cached locally in Cache Storage → prefetch, but grabbed from there so slow.

Perhaps there are many other open connections at the time this request was made? The browser will only execute so many connections concurrently before queuing new ones.

I was checking that, no other requests were present.

Just noticed even more weird thing: I see the same behavior on all recent starters we worked with inspecting them in Chrome. Also just tried our starter in Safari, and it seems works fine comparing to Chrome. The page response is instant from SW:

This has been logged as XDN-9494

For me the issue was resolved when Update on reload for the service worker was disabled in the application tab. We have it checked by default when doing development which means all the Service worker code needs to be run every time a page changes. After unchecking it 1.4s queueing time has gone down to 33ms which is in line with what I’d expect to see.