Using K3s’ Helm Controller to Deploy Charts via CRDs
Learn how K3s’ Helm Controller lets you declare Helm charts as Kubernetes resources, how to verify its operation, and what trade‑offs to consider for edge workloads.
ReadMeFeed / Community knowledge
Real questions. Useful conversations. Find the people who know your stack.
Learn how K3s’ Helm Controller lets you declare Helm charts as Kubernetes resources, how to verify its operation, and what trade‑offs to consider for edge workloads.
k3OS is archived, but its one-file config.yaml model — where hostname, SSH keys, and Kubernetes cluster membership are all declared at boot — is still a design worth stealing for edge and lab clusters.
K3s ships with an embedded SQLite database that removes the need for an external etcd, but the single‑writer lock and lack of replication limit its use to single‑node or low‑write workloads.
k3OS is archived, but its headline decision — an entire node defined by one versioned config.yaml — is still the right way to think about provisioning edge Kubernetes fleets.
Step‑by‑step guide to deploying a highly available K3s control plane using the built‑in SQLite (sqlitedb) datastore, covering bootstrap, verification, failure recovery, and migration to etcd.
Embedded SQLite makes k3s ultra‑lightweight, but write‑heavy workloads and shared storage can hit concurrency limits. Here’s how to evaluate the trade‑off and migrate to external etcd if needed.
Step‑by‑step guide to diagnose why a k3OS node boots but k3s fails to start, focusing on k3os-config validation and immutable filesystem checks.
When a k3s pod’s PVC stays in Pending, the culprit is often the local‑path provisioner. This guide walks through checking provisioner health, node disk space, stale PVs, and StorageClass config, with fixes tied to findings and escalation steps.
Learn how K3s uses an embedded SQLite database to replace etcd in single-node clusters, reducing resource overhead while maintaining full Kubernetes API functionality.