Some microservices fail to come into production due to lack of monitoring. Even with tests in controlled environments, it is very difficult to get a true picture of a microservice's performance. The production numbers will give us a final view of the scalability, availability, and performance of an application.
Not having alert systems and constantly collecting metrics from the microservices can cause a higher number of hits, override the application, or cause some instability in some segments of the system. Unmonitored instability is the most dangerous case because it is a silent error and difficult to detect; it will probably be too late by the time we find it.
In some cases, microservice instabilities happen because the development team did not understand the monitoring metrics of the application or did not know how to properly collect them.