Closed Traviskn closed 1 week ago
We wrapped our identify call in a helper that retries if it fails and we are now seeing a much lower failure rate
The identify method takes options and allows you to configure the timeout , {timeout: 30}
. Generally speaking indefinitely waiting for identify isn't a safe practice in case there is a network problem or outage.
The SDK should continue to retry even after the timeout, if it is not, then that is a bug and we need to look into it. We will look into it and validate.
Thank you, Ryan
There are upcoming connection handling improvements, but the ability of identify to timeout is as designed.
Describe the bug After upgrading to @launchdarkly/react-native-client-sdk version 10.2.0, we have seen around 12% of our iOS users get an identify timeout error. In contrast, only about 1% of our Android users are getting the timeout error. For some reason the iOS platform is seeing a much higher magnitude of identify timeout errors. We have a nearly equal number of active users between platforms, the sample size is not significantly different.
To reproduce Configure the launch darkly SDK and identify the user at app startup. Observe the rate of timeout errors, notice that the iOS platform is about 10% higher than the Android platform.
Expected behavior The identify call should time out at about the same rate between platforms.
Logs We utilize bugsnag for error tracking, here are the logs from a production error event:
SDK version @launchdarkly/react-native-client-sdk 10.2.0
OS/platform iOS