Duplicated crash reports: regular and WatchdogTermination

[kean]

Platform

iOS

Installed

CocoaPods

Version

8.9.2

Steps to Reproduce

In some scenarios, but not all, a single crash gets reported twice:

• Once with an actual stacktrace
• Once as an arbitrary WatchdogTermination

Expected Result

A crash results in a single even on Sentry. The crash rate is accurate.

Actual Result

The reported crash rate is inaccurate.

Occasion #1

Occasion #2

Are you willing to submit a PR?

No response

[kean]

The main issue is that you can no longer accurately tell the real app's crash rate.

It seems there is already logic that should take care of the duplicates, but it doesn't seem to work every time.

  // The app crashed on the previous run. No Watchdog Termination.
  if (self.crashAdapter.crashedLastLaunch) {
    return NO;
  }

I checked one of the events from the screenshots I provided, and they contain identical breadcrumbs and timestamps.

[kahest]

Hi @kean, thanks for the report! We'll investigate. Please note that due to current vacations and holidays we may be slower to response.

[brustolin]

Hello @kean. Is this a React Native project?

[kean]

Hey, @brustolin. Yes, parts of it are, and I think both of these crashes happened in the RN part.

[brustolin]

@krystofwoldrich do you have any idea what may cause this? Can you take a look?

[krystofwoldrich]

The WatchDog issue is created based on a heuristic, are you or some library call exit(0) when the fatal error happens?

Is it the WordPress-iOS app that you linked the issue to? Are you able to reproduce it locally?

[kahest]

@kean could you provide more information and ideally answer above questions? Does it still occur?

[kean]

The WatchDog issue is created based on a heuristic, are you or some library call exit(0) when the fatal error happens?

The crashes I originally reported were all coming the React Native part of the app.

it the wordpress-mobile/WordPress-iOS#20989 app that you linked the issue to? Are you able to reproduce it locally?

Yes, that's the correct repo, but I don't have a way to reproduce it.

This issue is currently much less of a concern as we managed to bring the overall number of the WatchdogTermination events significantly down. We suspect that disabling Sentry's app hang monitoring was the main reason why we saw the reduction, but it's hard to verify as there were many other changes that could've contributed to it.

[philipphofmann]

as we managed to bring the overall number of the WatchdogTermination events significantly down.

@kean, what did you do to achieve this?

We suspect that disabling Sentry's app hang monitoring was the main reason why we saw the reduction

Disabling app hang monitoring only stops the SDK from reporting app hangs, but the app hang tracker still runs to avoid tracking app hangs that get killed as watchdog terminations. I would be surprised if that would be the main reason.

[getsantry]

This issue has gone three weeks without activity. In another week, I will close it.

But! If you comment or otherwise update it, I will reset the clock, and if you remove the label Waiting for: Community, I will leave it alone ... forever!

---

"A weed is but an unloved flower." ― Ella Wheeler Wilcox 🥀