kubectl kedify plugin

The kubectl-kedify plugin provides a simple TUI-based interface for managing and troubleshooting Kedify installations.

Installation

Prerequisites

The plugin requires several dependencies to function properly:

macOS (Homebrew):

brew install bat curl figlet fzf kubecolor yq jq

Linux (apt):

apt-get install bat curl figlet fzf jq

Linux (yum):

yum install bat curl figlet fzf jq

For yq, consult the official installation guide.

Note: kubecolor is optional but recommended for enhanced output formatting.

Using Krew (Recommended)

Note

If you don’t have krew installed, first install it by following the krew installation guide.

Install the plugin using kubectl’s krew plugin manager:

kubectl krew install --manifest-url=https://github.com/kedify/kubectl-kedify/raw/main/.krew.yaml

After installation, verify it’s working:

kubectl kedify --version

Updating the Plugin

To update to the latest version:

kubectl krew uninstall kedify
kubectl krew install --manifest-url=https://github.com/kedify/kubectl-kedify/raw/main/.krew.yaml
kubectl kedify -v

Available Commands

The plugin provides several commands for managing Kedify:

• status - Show Kedify agent status and health
• logs - Display logs from Kedify components
• autoscale - Create HTTPScaledObject resources interactively
• debug - Low-level troubleshooting information
• insights - Analyze configurations for potential issues
• dump - Collect comprehensive diagnostic information

For detailed information about collecting diagnostic data, see Collecting Kedify Configuration.

KPA diagnostics

Compatible plugin builds resolve a ScaledObject’s generated name against both Kubernetes HPA and Kedify KPA. kubectl kedify debug scaledobject reports the actual kind and HPA-compatible KPA metrics, and warns when both kinds exist during a transition. kubectl kedify dump best-effort collects the KPA CRD and objects, controller resources and ownership-verified logs, events, and Prometheus metrics. Missing KPA CRD access is nonfatal, so the same commands continue to work on HPA-only clusters.

Source Code and Documentation

• Source Code: Browse the complete source at https://github.com/kedify/kubectl-kedify
• Issues and Features: Report issues or request features at https://github.com/kedify/kubectl-kedify/issues
• Releases: Download releases from https://github.com/kedify/kubectl-kedify/releases

Multi-cluster commands

Source: kubectl-kedify dispatcher and multicluster.sh, commit 0b12505ec7ea973aae7957d4a8f3895b9eebbc84. Check kubectl kedify --version and command help for the installed release. These commands manage distributed member access; they do not merely connect a dashboard fleet cluster.

Command	Arguments and behavior
kubectl kedify mc setup-member NAME	Required --keda-kubeconfig PATH, --member-kubeconfig PATH; optional --keda-context, --member-context, --member-api-url, --namespace (default keda), --yes. Creates member access resources and updates the central member kubeconfig Secret; inspect both contexts before confirming.
kubectl kedify mc list-members	--namespace, --keda-context, --output table|wide; inspect configured members without changing them.
kubectl kedify mc delete-member NAME	--namespace, --keda-context, optional --yes; removes central registration. Quiesce dependent distributed workloads first; do not assume it deletes all remote credentials/resources.
kubectl kedify mc help	Displays the available subcommands.

Register members with GitOps is the declarative alternative; distributed scaling explains the topology.

Other command scope and errors

The dispatcher also provides install/i, delete/d, status/s, logs/l, autoscale/a (including ingress mode), debug/dbg, insights/ins, dump/dmp, version and help. Install/delete/autoscale can mutate the selected cluster; confirm the cluster context and the component’s Helm/GitOps owner before those operations.

The plugin’s insights command analyzes ScaledObject configuration, including polling/fallback choices. It is not hosted Insights resource recommendations. The documented HTTP workflow authors ScaledObjects; legacy interactive autoscale behavior must be checked against the installed plugin before using it.

For missing commands, check plugin discovery, PATH and version; for dependency errors, install the named prerequisite. For permission failures, inspect the selected context/namespace and required resource access rather than rerunning as an unbounded administrator. Use diagnostic collection and review bundles for credentials before sharing.