问题

WE currently use flurry for our iOS app analytics.. we've discovered a recurrent phenomenon every time we submit a new version of our app to the app store.. whenever we do that, we see a spike in the user activity graph on Flurry:

We have our build configurations made so that no development testing ever make it to the prod app on flurry (we created a dev version for testing).. this is what we have in our - (BOOL)application:(UIApplication *)application didFinishLaunchingWithOptions:(nullable NSDictionary *)launchOptions method:

#ifdef RELEASE
    NSLog(@"launching app in RELEASE 
    [Flurry startSession:@"RELEASE-APP-KEY"];
#else
    NSLog(@"launching app in NON-RELEASE mode");
    [Flurry startSession:@"NON-RELEASE-APP-KEY"];
#endif

and our tests ensure that launching app in NON-RELEASE mode shows up on Xcode's console whenever they're doing testing..

further.. as soon as the app launches.. we also have inside the didFinishLaunchingWithOptions method a call to our server that checks the app version (ie /app/ios/version_state).. going over all our logs during the spike period in flurry.. we couldn't find a single request with the version flurry is showing the spike in..

Any idea what this may be all about?

hypothesis

could it be that flurry is doing some weird caching behavior or that whenever an is uploaded to apple store.. apple runs some automatic tests that checks if the app launches at all?

回答1:

First possibility: those users are related to Apple testing your update. They do so every time you send an update. Theoretically , it's a manual, non-automated test. But nothing would prevent them to introduce some automatic launches. In that case: You should check whether the devices for those users are different in Flurry. If the above doesn't apply, I would assume one of two possibilities: either Flurry is wrong or you are wrong. If we wanted to deep digger, it would be necessary to know: Are those spikes related to the latest version? And also: if that small amount of users is important to you (I mean, there are apps intended to reach 10 to 20 people and apps that target millions of users), then I would move to a paid real-time service such as adjust, appsflyer... You could choose one that suits you.

回答2:

I work with the Flurry Integration team. I would check if you have background sessions on setBackgroundSessionEnabled.

What might be happening is that as all of your users auto-update, a background session is being kicked off, causing the spike you're seeing on the new version.

来源：https://stackoverflow.com/questions/33145361/how-to-prevent-spike-in-flurry-user-activity-every-time-a-new-version-is-pushed

标签

ios