# Missing Feedback Push Notifications

## Timeline
**7/13**: merged in change from 15 to 45 minute delay (`FEEDBACK_PUSH_NOTIFICATION_BUFFER`) in `Order` model.
**7/14 - 7/16**: used wrong forms => no `Viewed Feedback Prompt` event was fired.
**7/17**: ran a new rake task to change feedback form settings, and the`Viewed Feedback Prompt` numbers start coming back up. But not to previous levels: [Amplitude chart](https://analytics.amplitude.com/slice/chart/v8fe50b).
**7/19**: sharp drop in jobs getting scheduled: [DataDog dashboard](https://app.datadoghq.com/dashboard/g49-eux-42k/feedback-push-notifications?from_ts=1593366323412&live=true&to_ts=1595958323412).

## Investigation
At first I assumed the drop in the `Viewed Feedback Prompt` numbers were due to the changes I had made to the feedback forms.


But when I created the [DataDog dashboard](https://app.datadoghq.com/dashboard/g49-eux-42k/feedback-push-notifications?from_ts=1593366323412&live=true&to_ts=1595958323412) I realized that I might be wrong about that.

All the job metrics in that dashboard come from jobs scheduled in the [Order::ensure_payment_for](https://github.com/slicelife/myr-admin/blob/c7863f33e3656c5ac686149353e9b15f9348eb2d/app/models/order.rb#L2240) method. I figured it would make sense to see if the job was still getting scheduled at the same rate, which is why I created the DD dashboard.


When I saw the drop in number of times it gets scheduled, I looked at metrics for the other jobs also scheduled at the same time and that's when I saw the drop in all numbers.


I assume this means that it's `ensure_payment_for` that is getting called less. But I have no idea why. I tried looking for a PR around that time that could have caused this issue, but I lack the admin experiemce to be very effective :-D

Since it’s a drop in all the jobs scheduled in the `ensure_payment_for` method I’m not entirely sure it can be explained by my changing the wait time on one individual job. I'm also wonderign if this is due to a change in how metrics get recorded...

Looking at sidekiq [logs](https://app.datadoghq.com/dashboard/hcx-tbx-5if/ops-worker-dashboard?from_ts=1593373999207&fullscreen_end_ts=1595967079647&fullscreen_paused=false&fullscreen_section=overview&fullscreen_start_ts=1588191079647&fullscreen_widget=4685503894683312&live=true&to_ts=1595965999207), things seem steady: