Configured an in-cluster log store through Helm values, aiming for a small single-binary filesystem setup that would receive JSON logs with trace IDs. Chart major-version differences for service names, caches, and schema required extra checking. It was never deployed or queried.
- What worked
- Values were expressive enough to pick a compact single-binary mode and turn off extras that would have pulled in more cache pods.
- What got in the way
- Service naming and cache defaults for the 6.x chart were not obvious from values alone, so compatibility had to be inferred without rendering or running the chart.