After updating at my work Kubernetes cluster and updating Prometheus Operator, was updated its component – kube-state-metrics to 2.3.0. Early when versions were used 1.4.0 and 1.7.0, there is no problem because it is back compatible. But since 2.0.0 was deleted some metric. And I need to discover this, because some of them was used in alerts!
I've met a problem with using SQLAlchemy. An error in commit() cause the call to get a new sequence of the table primary key. It updates the sequence number. But rollback() doesn't fix it.
Everytime when I am doing a something by Python (or by something else) I need to use articles about main libraries. Despite that I've seen it a lot of times, it just can't fit in my head. So I wrote this article to get the information quickly.