Currently our primary app does log some error events to mixpanel but from my understanding we haven't historically used them much for taking action such as knowing a bad release went out if error rate spikes. While this would be a larger task to start getting better error reporting out of the main application, memoless presents an app with 1 critical flow that we might be able to experiment with some better error reporting and alerting so we aren't so reliant on users reporting bugs as the primary trigger to roll back a release.
Mixpanel also does offer custom alerts that could be monitored by operations.
Acceptance Criteria
add mix panel event triggers for event triggers or anticipated bad states to the new og app
determine how best to get decent information out of these events that help with debug-ability / traceability
create a mixpanel dashboard to track these errors
create an alert with the assistance of operations that can be monitored by them
The primary goal of this would be to move towards less reliance on users and being more proactive when a bad release goes out the door hopefully reducing the time it takes us to get a fix out (mttr type metric, which we don't actually track, yet)
0xean
commented
5 months ago
Background
Currently our primary app does log some error events to mixpanel but from my understanding we haven't historically used them much for taking action such as knowing a bad release went out if error rate spikes. While this would be a larger task to start getting better error reporting out of the main application, memoless presents an app with 1 critical flow that we might be able to experiment with some better error reporting and alerting so we aren't so reliant on users reporting bugs as the primary trigger to roll back a release.
Mixpanel also does offer custom alerts that could be monitored by operations.
Acceptance Criteria
The primary goal of this would be to move towards less reliance on users and being more proactive when a bad release goes out the door hopefully reducing the time it takes us to get a fix out (mttr type metric, which we don't actually track, yet)