How to categorize this issue?

/area monitoring
area control-plane
/kind enhancement

What would you like to be added:

At present the metrics exposed by druid regarding compaction jobs are very minimal, i.e it categorises the jobs into just succeeded or failed, but in reality the failed category is vast, it can fail for multitude of reasons with the prominent reasons being

1. Process Failure
2. Deadline Exceeded
3. Preemption
4. Eviction

There can be multiple sub reasons for Eviction like DeletionByTaintManager, EvictionByEvictionAPI, TerminationByKubelet, etc

The reason for why knowing about the failure reason matters is because currently we expose these compaction metrics from druid which gardener project scrapes through prometheus for monitoring and alerting, and we've been seeing a lot of jobs tagged under failed category but upon monitoring for some time we've seen that these are not because of the actual process failures, but due to external disruptions and job exceeding the deadline most often. So it's important that we expose the actual reason for the failure instead of just saying it's failed. This way, it becomes extremely clear as to what is causing the failure and the actions that can be taken to minimise such failures.

This is important not just for gardener project, but for external independent consumers of etcd-druid as well as they can scrape the prometheus metrics exposed out of etcd-druid for the compactions job if they've enabled them to better monitor and understand the compaction job lifecycles.

Why is this needed:

To improve monitoring of the compaction jobs to better understand the reasons for job failures so that necessary targeted action can be taken to minimise such failures.