Environment

SaaS (https://sentry.io/)

Steps to Reproduce

1. Install sentry on any angular application and run it in production
2. Ensure tracing is set to 100% so you capture everything
3. Ensure you have Trace IDs being sent in http requests so you can verify that it is exactly the same one you are looking at in sentry
4. Go to that application website and open the network tab - i used chrome and firefox to verify it happens regardless of browser
5. Find the same page load in sentry and compare the times chromes network captured timing vs sentry

Expected Result

The timings should be nearly identical, if not somewhat close - if chrome network said a request took 125ms, I'd expect sentry to be at least close to that.

Actual Result

Sentry cites times 5x higher than any other network inspector. Each http timing metric is multiple times higher than any other tool checking network traffic timings I've found, making timing metrics from Sentry unreliable. I built a dashboard in Sentry to measure our apps timings from various regions and realized that even when I filter to a single transaction, that Sentry cites each http request as nearly 5x longer timing than both Checkly automation logging and manually checking requests in Chrome on a cache-free and incognito session browser were citing.

Attached is a screenshot of just one http request as a sample where the timing diverged, but it happens on every single http request - this is captured in the middle of a page load transaction, which did not get closed early or otherwise interrupted in any way, and got a 200 response. The timing difference is 5.6x higher than the reality in the Network inspector.

Left is sentry, Right is the Network Tab

Product Area

Performance

Link

No response

DSN

No response

Version

No response