I am a developer at a company that hosts a single page web application written in Angular. We are experiencing an issue when our customer base uses iCloud private relay in Safari on iOS / Safari 16.4+. The issue is most pronounced at a point that is critical to our conversion process, and we are losing a lot of customers because of it. The customer is able to load our Angular app and start the account creation process, but at some point shortly after creating their account, all network requests (including those to third parties such as launch darkly, okta, fullstory, and grafana) begin to fail within the browser. We attempt to retry the requests when they fail after they've met our timeout interval, but the requests still fail to produce responses. The discrepancy between the network response time on the calls just prior to the requests hanging seems to indicate stable networks that are producing sub 200ms response times and then all subsequent requests don't respond. Using some analysis tooling, we've been able to identify that 100% of the use cases in which this issue occurs have traffic proxying through iCloud Private Relay. Below are the top 20 IP addresses that have resulted in this issue since May 1st. This issue was not present prior to iOS 16.4 release and has yet to improve. 104.28.48.216 104.28.39.142 146.75.248.0 104.28.151.113 104.28.32.93 140.248.30.0 104.28.48.217 104.28.103.19 104.28.103.21 104.28.97.24 146.75.164.1 104.28.97.17 104.28.39.146 104.28.103.23 104.28.97.25 104.28.39.141 140.248.0.1 104.28.97.21 104.28.97.22 104.28.132.126
As an additional note, we have finally figured out a solution to prevent this bug from happening. We unregistered our Angular service worker if the browser was Safari and this behavior stopped occurring.
Thank you for the report! Could you please file it via https://feedbackassistant.apple.com with a sysdiagnose taken after this reproduces? This will likely require collaboration between multiple Apple teams, which cannot take place in WebKit Bugzilla.