A GitOps rollback needs time to reach Git
A live rollback fixed a crashlooping workload, but the more interesting part was needing to suspend reconciliation long enough for the fix to become the desired state instead of a temporary lie.
Writing
Everything stays flat and chronological. If the archive is short, that is an honest reflection of how often something is worth publishing.
A live rollback fixed a crashlooping workload, but the more interesting part was needing to suspend reconciliation long enough for the fix to become the desired state instead of a temporary lie.
The monitoring split that held up best was also the less ambitious one: keep Gatus on in-cluster checks, keep edge probes in blackbox-exporter, and stop trying to make one checker own every route shape.
Homelab analytics had enough auth surface to accidentally grow a second identity product. The useful decision was to keep identity proof upstream, keep authorization local, and demote local login to a narrow break-glass path.
Backend-owned contracts only become useful release artifacts once the compatibility tooling is conservative enough to complain about real breakage instead of politely missing it.
The interesting part of the remote dev shell cutover was not SSH or code-server. It was moving the real working state onto shared storage without corrupting repo-local databases or pretending two writable trees could coexist peacefully.
The sprintctl and kctl workflow is useful largely because it does not pretend every intermediate agent artifact belongs in Git. Live state stays local, committed artifacts stay small, and promotion has to be deliberate.
Backend-owned contract exports are mainly a way to stop arguing with a stale frontend view of reality. The useful part is not code generation by itself but making the backend the source of truth and then checking the drift explicitly.
The first post is mostly a marker that the site exists and the machinery got here before the archive did.