feat: add kubernetes-agent solution (CX-41) - #241
Merged
Merged
Conversation
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…aths setup.py now prompts the user to choose between: - Spinning up a GitHub Codespace with a kind cluster automatically (via gh codespace create + gh codespace ssh for all k8s/helm commands) - Deploying to any existing Kubernetes cluster configured in kubectl context (runs kubectl/helm locally as before) Idempotent: Codespace name is persisted in state so re-runs reconnect rather than creating a new Codespace. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…despace wait message
…eflect longer onCreate
Previously the existing-Codespace path returned immediately after verifying the Codespace exists, without waiting for kubectl/kind to be ready. Now both the new and reuse paths converge on the same kubectl readiness poll. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Tee all onCreate.sh output to /tmp/onCreate.log for diagnosability - Use python3 -m pip instead of pip (more reliable on Ubuntu 24.04) - Timeout error now shows the SSH + log command to diagnose failures Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- onCreate.sh writes /tmp/onCreate.failed on non-zero exit (EXIT trap) - Poll loop checks failure sentinel on each iteration via a single SSH call, failing fast instead of waiting out the full 15-min timeout - Combined kubectl check + sentinel check into one SSH round-trip - Corrected wait message to "may take 15-20 min" (was "~5-10 min") Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The CLI runs locally; installing it in the Codespace is unnecessary and was causing onCreate.sh to fail (Ubuntu 24.04 base has no python3 in PATH by default), which triggered Alpine recovery mode. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Both the FAILED sentinel and timeout error paths now SSH in and fetch /tmp/onCreate.log automatically, printing it inline so the user sees the failure reason immediately without manual steps. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…cess - Phase 1 (Codespace Available) and Phase 2 (kubectl ready) now each get their own full CODESPACE_READY_TIMEOUT budget instead of sharing one — Phase 1 taking 5+ min was consuming kubectl polling time - Add sshd devcontainer feature so SSH is available immediately when the Codespace starts, before onCreateCommand completes Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
…emote Non-interactive SSH sessions don't load .bashrc, so kubectl doesn't find ~/.kube/config via KUBECONFIG. The readiness poll now uses the explicit binary path and kubeconfig. _run_remote uses bash -lc (login shell) so PATH and env are set correctly for kubectl/helm calls. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
gh codespace ssh joins post-'--' args with spaces before the remote shell sees them, so shell metacharacters (;, >, |) in our bash -c script were interpreted by the remote shell, not bash. Replace with two simple SSH calls: 'test -f /tmp/onCreate.failed' for the sentinel and direct kubectl invocation for readiness — no shell metacharacters. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
gh codespace ssh joins post-'--' args with spaces before the remote shell sees them, making it impossible to safely pass scripts with shell metacharacters (|, >, ;). bash -lc also sources profile scripts that can corrupt stdout of piped commands (e.g. kubectl apply -f -). _run_remote now writes the script to a local temp file, copies it to the Codespace via gh codespace cp, then executes it with a clean env. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
gh codespace cp wraps remote paths in single quotes (scp behavior), making the destination filename include literal quote characters. Instead, pipe the script to 'tee /home/vscode/cortex-run.sh' via SSH stdin — no file path quoting issues, no shell metacharacters. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
The helm install script installs with -rwxr-xr-- (750), which the vscode user (non-root, not in root group) cannot execute. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Client-side kubectl apply stores the full manifest in a last-applied-configuration annotation, which exceeds the 262144-byte limit for the large Argo Rollouts CRD. Server-side apply avoids this. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
When metadata labels are configured, Cortex bypasses cortex.io/tag annotation lookup entirely — the two strategies are mutually exclusive. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Data Model: ASCII art renders in the CLI 'What next?' menu - Architecture -> Data Model (matches _extract_first_codeblock path) - Next steps -> After Installing (matches _extract_section lookup) - Updated After Installing content per product guidance Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
kubernetes-agentCortex solution bundle, which deploys the Cortex k8s-agent to a Kubernetes cluster and creates demo workloads to demonstrate the integrationonCreate.shWhat's included
cortexapps_cli/solutions/kubernetes-agent/setup.py— post-install setup scriptcortexapps_cli/solutions/kubernetes-agent/helm-chart/— bundled k8s-agent Helm chartcortexapps_cli/solutions/kubernetes-agent/manifests/— demo workloads (Deployment, StatefulSet, CronJob, Argo Rollout)cortexapps_cli/solutions/kubernetes-agent/catalog/— demo-kubernetes Cortex entitycortexapps_cli/solutions/kubernetes-agent/README.md— with ASCII art data model and After Installing next steps.devcontainer/kubernetes-agent/— devcontainer config andonCreate.shTest plan
What next?menu: Data Model (ASCII art) and Next Steps (After Installing) both render correctly🤖 Generated with Claude Code