Improving observability when marking packages as pending_destruction

🔥 Problem

This part is silencing errors which makes hard to debug issues such as this one.

Why this part was implemented like this?

This part is doing 2 very simply operations:

1. Update the status of the given package to mark it as pending_destruction
2. Enqueue a job to mark all package files of this package as pending_destruction

It's strange to have an error with such simple operations.

🚒 Solution

Catch the error object and track it in the log system so that we know what is happening.