Starting about Feb 15, we’ve noticed 1 out of 10 fetch calls take “too long” anywhere from 10-30 seconds when a typical call should take 5-10. Our app is critically dependent on having an initial fetch call on app startup, and secondarily having a continual repeating fetch cycle to keep real-time data fresh.
This is observed in both SDK 31 and 32 on both platforms, and most affecting Android.
We have done local testing to eliminate as many variables in terms of what could delay a call and tested with other mock http endpoints. In addition we’ve mocked up native iOS and Android apps to verify our back-end endpoints are not the reason.
As far as we can tell there is something that changed under the hood in the past 2-4 weeks that sometimes gives a long delay to a fetch call. It’s almost like a fetch is no longer given priority on the thread.
This also coincides with our Analytics suddenly recording Expo line errors {} in our fetch catches which in the past have never occurred.
I see the last ExpoKit updates were 30 and 15 days ago but the changes don’t necessarily seem related to the data handling portion. I wonder, is there some cloud or back-end infrastructure change done to Expo servers that may cause this odd delay? I’m not familiar how the bridge to native datatask works but I suppose the only reason is if the data calls rely on a proxy or there is some interaction with the Expo servers.
It could also be, this slowness is just being more closely observed now and that it relates to something that changed to React Native from 30 to 31. Our next test is to move critical fetch calls to different parts of the app or sooner in the component tree. Right now we run fetch calls from separate non-component helper modules, maybe RN de-prioritizes them for some reason.