Node - Seeing inconsistent results in the dashboard

Using @sentry/node and captureMessage, I’m seeing inconsistent results in the dashboard.

I have a function that is called 5 times, 1 minute apart. Each time it sends something like Sentry.captureMessage({messageId} - {attempt})

If I run this several times, I’ll see several messages in the dashboard, like:
abc - 2
xyz - 1
rgb - 3

The attempt numbers do not match the number of events in the Sentry dashboard. For example, ‘rgb - 3’ was the third event sent to Sentry, but Sentry says there is only 1 event for that message.

I can see the full aws logs to convince myself the function ran as expected, but the Sentry dashboard does not consistently update. I don’t know if it’s deduping some messages but not others, or where the inconsistency is coming from.

I’m open to user error, so please, any suggestions are welcome.

I started a related github issue here with more info here. The issue is asking if captureMessage should use the scope fingerprint, which it does not seem to do. But this inconsistency seems like it may be a separate issue.

I tried setting sampleRate: 1 in Sentry.init but it made no difference. I just sent 6 messages and after ten minutes, the dashboard only shows two: #4 and #6. The other 4 messages never showed up.

I also switched from captureMessage to captureEvent({ message: myMessage }) because captureMessage does not seem to respect the scope.fingerprint, according to the dashboard section for EVENT GROUPING INFORMATION.

I’ll attach a screenshot of my custom fingerprint just in case, but it’s constant for all of these calls.

Screen Shot 2020-09-23 at 4.06.14 PM

Any ideas? This needs to be consistent.

Sentry support suggested using a separate key to eliminate the possibility of other services causing this one to hit the rate limit. I’m setting the sampleRate to 1 during Sentry init.

I also started logging the response from Sentry.captureEvent, which seems to be an eventId created by Sentry. My logs confirm that all of my calls are returning ids, but not all of the ids show up in the Sentry dashboard.

My conclusion so far is that Sentry is either deduping or dropping these events.