Self-Recovery Subflow with Kestra KV Store

Automatically purge expired key-value entries from Kestra's internal KV store across namespaces on a daily schedule.

Cache LLM answers in Redis so repeated questions are served instantly and OpenAI is only called on a genuine cache miss.

Run scheduled tasks only on business days using a Python calendar check that skips weekends and country-specific public holidays.

A deliberately failing flow that demonstrates block level errors, flow level errors, automatic retries, and KV context for alerting flows.

Run a Python script inside a Kubernetes pod with declarative CPU and memory limits, straight from a Kestra flow.

Collect logs from every pod in a Kubernetes namespace within a time window, plus namespace events, in one run.

Reset a Flaky System from a Self-Healing Recovery Loop