install_barman_plugin() fetches from GitHub on every run (up to 6×120 s
retries), stalling the installer worker thread even when the CNPG cluster
is already at 'Cluster in healthy state' with all pods 2/2 Ready.
Add a fast-path at the top of initialize(): check .status.phase for
'healthy' then verify all pods show 2/N ready — if both pass, return
immediately. Full pipeline (ensure_operator, install_barman_plugin,
_apply_manifest, _wait_cnpg_pods) is only entered when needed.
Also adds test_cnpg_initialize_skips_when_cluster_healthy to
test_deployment_mode_isolation.py to guard against regression.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
kube-prometheus-stack CRD re-application stalls on every installer run
even when the release is in deployed state with all pods Running. For k3d
this is purely idempotency noise — skip the upgrade and return immediately.
Broken state (failed status, unbound PVCs) still triggers a full purge+reinstall.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
k3d registry containers are addressed as k3d-NAME.localhost:PORT from
inside the cluster, but containerd resolves that as 127.0.0.1 (loopback)
which has nothing listening. The fix is to write /etc/rancher/k3s/registries.yaml
on every server/agent node to mirror k3d-NAME.localhost:PORT -> http://k3d-NAME:PORT
(Docker network DNS), then restart the node so containerd picks it up.
_patch_k3d_registry_mirrors() is called from _ensure_k3d_registry() after
the registry is confirmed running (idempotent — skips nodes already configured).
This makes knoe-authority and any other image pushed to localhost:5000
immediately pullable inside the cluster as k3d-knoe-registry.localhost:5000/...
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Three issues caused the purge-and-reinstall loop:
1. _purge_broken_monitoring deleted ALL PVCs including Bound ones; grafana's
working PVC was wiped on every retry. Now only delete PVCs in Pending state.
2. Broken-state detection keyed on Pending pods + unbound PVCs, which is true
during any still-converging install (including ones that timed out but
whose pods eventually came up). Gate on helm status=='failed' + unbound PVCs.
3. k3d install used --wait, which blocks on all kube-prometheus-stack components
(prometheus, alertmanager, node-exporter). They converge async after the
Prometheus operator starts; --wait always timed out. Drop --wait for k3d;
the status_common_services.sh check verifies readiness independently.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
kube-prometheus-stack pulls multiple large images on a fresh k3d install;
600s / 10m is not enough. Set --timeout 20m and process timeout 1320s
(20m + 2m buffer) for k3d, keep 10m for k8s/k3s where images are warm.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Two bugs kept prometheus-grafana-0 Pending for 21 days in k3d:
1. update() routed k3d through _values_yaml_k3s() which hardcodes
merlin-local-iscsi-* storage classes and merlin.prole.org node affinity
— neither exists in k3d, so grafana's PVC could never bind.
Add _values_yaml_k3d() that uses local-path (or MONITORING_STORAGE_CLASS
override) with no node affinity; dispatch k3d separately in update().
2. No broken-state recovery for non-k8s mode. Add _purge_broken_monitoring()
(helm uninstall + delete stuck PVCs) and call it from update() when
Pending pods and unbound PVCs are both detected in the monitoring namespace.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
k3d-prole-registry was squatting on port 5000, leaving k3d-knoe-registry
in created state. Two bugs made this invisible:
- k3d_registry.py: substring match (name in stdout) matched k3d-knoe-registry
as already-exists without checking STATUS=running; add _k3d_registry_running()
requiring last column == running, detect and delete prole-registry before
creating knoe-registry, recreate if found in non-running state
- status_common_services.sh: grep -qx 'knoe-registry' (exact) never matched
k3d prefix 'k3d-knoe-registry'; fix with awk suffix match + STATUS==running
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Junie's session targeted the prompt "stabilize ./install.py -c conf/k3d.cfg
using strict TDD" — broad installer-side work, not the k3d-mirror Phase 3
brief I had filed (which she didn't pick up; phase-3 brief stays open). All
750 installer tests pass post-change.
What Junie produced:
install.py (NEW) Top-level CLI entry point. Was
imagined by the prompt but didn't
exist; this commit makes it real.
knoe/deployment.py (NEW) `KnoeDeployment` orchestrator for
the k3s service-mode deploy pipeline.
Wraps Ansible kubeconfig fetch,
opentofu apply, init_*.sh post-apply
scripts, and (optionally) supabase/
deploy.sh.
knoe/ui/screens/cluster.py Dual-cluster GKE kubecontext UI: prod env
knoe/ui/screens/cfg.py now shows separate "App Cluster:" and
"DB Cluster:" dropdowns instead of a
single "Kubernetes Context:" combo.
New _app_kubectx_combo + _db_kubectx_combo
widgets; new app/db_cluster_kubecontext
tk.StringVars.
knoe/core/{actions,env,milestones}.py
knoe/core/ops/storage.py
knoe/config.py, knoe/knoe_conf.py Plumbing changes for the dual-cluster
kubecontext flow + storage-class topology
detection cleanup.
knoe/tools/cleanup_cnpg_storage.py (NEW) Stand-alone cleanup utility.
tools/dashboard.sh (NEW) Dashboard helper.
conf/knoe.cfg (NEW) Master cfg generated by knoe_conf.
conf/dev/ (NEW) Dev-mode cfg directory.
conf/port-mapping.cfg Port mapping tweaks for k3d.
tests/installer/* (8 files) New + extended tests for the dual-cluster
tests/test_database_options.py TUI, kubecontext save flow, storage ops,
topology detection, deploy helpers,
database-options screen.
Issues found in Junie's working state and fixed here:
1. install.py was a 11-line import shim with no shebang, no `chmod +x`,
no `if __name__ == '__main__'` block. `./install.py -c conf/k3d.cfg`
returned `Permission denied` and `python install.py` did nothing.
Added `#!/usr/bin/env python3`, `chmod +x`, and a __main__ block
that delegates to `knoe.ui.screens.main()`. `./install.py --help`
now prints the canonical argparse help.
2. knoe/deployment.py had FIVE `subprocess.run()` call sites with no
`timeout=` argument (`_run_script`, `_run_cmd`, the Ansible playbook
fetch, `tofu init`, `tofu apply`). A hung child process — typical
failure mode is a script waiting on stdin or a stalled network
call — would lock up the installer indefinitely. Added timeouts:
- Ansible kubeconfig fetch: 120s
- tofu init: 300s
- tofu apply, _run_script, _run_cmd: bounded by new module
constant `_MILESTONE_TIMEOUT` (default 1800s = 30 min, override
via `KNOE_MILESTONE_TIMEOUT_SECONDS` env var).
`subprocess.TimeoutExpired` is caught explicitly; on timeout the
run helpers return exit code 124 (conventional timeout code).
3. `conf/k3d.cfg` was corrupted with MagicMock string-reprs on disk:
KNOE_CONF = <MagicMock name='Canvas().tk.call().strip()' id='4743999712'>
argocd.node_selector = <MagicMock name='mock.StringVar().get().strip()' id='...'>
Likely path: Junie ran `./install.py -c conf/k3d.cfg` interactively
in a non-Tk environment (or with a partially-mocked widget set) and
the installer's "save current state" path wrote the mock-objects'
`__repr__` strings into the cfg file. This commit reverts the cfg
to its pre-Junie state. **Followup: harden the cfg save path
against non-string widget values** — track separately.
4. The corrupted cfg caused the installer to call `os.makedirs()` on
the mock-string values, producing 10 directories on disk literally
named `<MagicMock name='Canvas().tk.call().strip()' id='4733210304'>/`
etc., with 5–86 files of install artifacts inside each. Removed.
The "final step is timing out" the user reported was almost certainly
issue #2 above: install.py walked the milestone pipeline, hit one of
the unbounded subprocess.run calls, and the wrapped command (probably
supabase/deploy.sh, which Junie was reading for context when her
session timed out) hung. With the timeouts in place that path now
exits cleanly with rc=124 instead of locking up.
Verification:
- pytest tests/installer/ -q 750 passed in ~25s
- python3 -c "import knoe.deployment" imports clean
- ./install.py --help prints argparse help
- find . -maxdepth 1 -type d -name '<MagicMock*' | wc -l 0
- head -7 conf/k3d.cfg clean (no MagicMock)
Out of scope for this commit (followups):
- The cfg save-path that wrote mock-objects-as-strings (issue #3 root cause).
Reproducer: launch the installer in an env where Tk widget vars are
`unittest.mock.MagicMock` instances. The cfg save code should refuse to
serialize non-str values rather than calling `str()` on a MagicMock.
- The k3d-mirror Phase 3 brief (`docs/plans/junie/k3d-knoe-auth-pod-deploy.md`)
is still open — Junie picked a different prompt this round.
Co-authored-by: Junie <junie@jetbrains.com>
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- Introduced `_recover_stale_prometheus_helm_lock` to detect and recover from stale Helm release locks during Prometheus deployment.
- Enhanced Prometheus deployment flow with stability verification and rollback to the last deployed state.
Summarize CLAUDE.md updates:
- Document both live GKE clusters as e2-standard-2 x3 in us-west3 and note stale app default in code.
- Call out DB cluster naming mismatch between conf (knoe-dev-cnpg-0) and code default (knoe-cnpg-0).
- Add SSD quota guidance: 300GB fully consumed by CNPG, so non-CNPG PVCs must use standard/pd-standard.
- Refresh reset script reference and quota section details to match current runtime state.
Co-authored-by: Junie <junie@jetbrains.com>
- Added fallback logic to resolve Garage ILB hostname if IP is unavailable.
- Improved handling of `self.inputs` with prioritized merging for better automation.
- Updated config persistence to use `write_config` for milestones and repairs.
- Extended Garage ILB IP retry logic to 5 minutes for GKE provisioning delays.
- Changed ILB deployment context to DB cluster (`knoe-cnpg-0`) for GitLab object storage setup.
- Renamed service to `garage-s3-ilb` and updated related references.
- Enhanced logging for Garage ILB IP resolution with retry and error handling.
- Persisted resolved ILB endpoint to config files for improved reliability.
- Refined split-cluster Garage deployment workflows with clearer context management and debugging.
- Introduced `get_endpoint_ip` utility to fetch Garage ILB IPs across k3s, k3d, and k8s modes.
- Updated initialization to deploy Garage on the DB cluster when using split GKE clusters.
- Enhanced logic to persist resolved Garage ILB endpoints for GitLab object storage configurations.
- Added Redis master resolution and removed placeholder Garage endpoints during deployment.
- Updated post-init scripts to include Redis initialization alongside Kong and backup scripts.
Switch production config to k8s/GKE contexts and align service naming. Add immutable StatefulSet update fallback for Garage across k3d/k3s/k8s. Harden CNPG deploy and backup bootstrap paths, and update installer coverage for CNPG webhook and Garage common ops.
Co-authored-by: Junie <junie@jetbrains.com>
KNOE_IMAGE_REGISTRY auto-detection in cloudnative_pg._apply_manifest():
- Explicit config wins; falls back to gitlab-registry when GitLab ns exists;
final fallback to registry:2 in knoe-system. No longer raises when unset.
conf/service/prole.cfg: point KNOE_IMAGE_REGISTRY at gitlab-registry since
GitLab is already deployed on this cluster.
etc/init_registry.sh: add 'migrate' action that copies all images from
registry:2 to gitlab-registry via skopeo (prints commands if skopeo absent).
Run after GitLab is up, before decommissioning registry:2.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add [CNPG Clusters] section to prole.cfg as the authoritative source for
CNPG cluster definitions. The deploy pipeline reads this registry instead
of computing version-appended namespace names at runtime.
- conf/service/prole.cfg: add [CNPG Clusters] with knoe-db as identity
cluster; fix DATABASE_NAMESPACE=knoe-db (was knoe-db-18-140); add
KNOE_IMAGE_REGISTRY for the k3s in-cluster registry hostname
- deploy/opentofu/k3s/manifests/prole/knoe-db.yaml: replace hardcoded
registry.test-system image with templated KNOE_IMAGE_REGISTRY
- knoe/core/actions.py: add _cnpg_cluster_registry(), _cnpg_identity_cluster();
fix _initial_db_namespace() to use declared namespace without version suffix
- knoe/core/milestones.py: iterate over all declared clusters in Init and
Deploy milestones; per-cluster CNPG_IMAGE_NAME support
- knoe/core/ops/cloudnative_pg.py: resolve KNOE_IMAGE_REGISTRY and
CNPG_IMAGE_NAME overrides in _apply_manifest()
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- separate default app and db cluster contexts (knoe-dev-0 and knoe-cnpg-0) across actions, ops flows, and installer config\n- update CNPG placement and GKE manifests to keep PostgreSQL on the db cluster while app services stay on the app cluster\n- add GCS-backed Barman ObjectStore manifest and cross-cluster Garage patch workflow\n- refresh UI cluster/database/service screens and monitoring wiring for dual-cluster operation\n- add reset/patch scripts, db context selection test coverage, and architecture/network documentation updates
Co-authored-by: Junie <junie@jetbrains.com>
- harden CNPG backup ObjectStore readiness fallback in statusless clusters by validating controller availability and expected spec
- improve backup/plugin reconciliation paths and related test coverage for barman-cloud flows
- include ongoing GKE/CNPG deployment and configuration updates from this checkpoint
Co-authored-by: Junie <junie@jetbrains.com>
UI screens
- database.py: fix mode detection to use env_key priority (prod→k8s, service→k3s) so stale DEPLOYMENT_MODE never overrides the user's chosen environment
- database.py: Registry status reads ARTIFACT_REGISTRY_AVAILABLE persisted by cluster screen; uses SERVICE_NAMESPACE for Artifact Registry repo name
- cluster.py: add Artifact Registry traffic light (amber→green/red) to prod section; _check_artifact_registry_async persists ARTIFACT_REGISTRY_AVAILABLE into Global cfg
- cluster.py: re-trigger Artifact Registry check after GKE cluster selection so the light re-evaluates once region is available from KUBECONTEXT
- cluster_nodes.py: fix TclError on Python 3.14 — pady=(2,0) tuple → pady=2 scalar
- __init__.py: seed knoe-system namespace when saved value is "default", not only when empty
- services.py: replace hardcoded "Prole DB" log string with dynamic cnpg_cluster name
Core ops
- cloudnative_pg.py: replace one-shot Barman plugin retry with 6-attempt loop; first cert-manager/x509 failure triggers rollout restart + 30 s CA propagation wait; subsequent failures back off up to 60 s per attempt
- cloudnative_pg.py: TLS CA CN now uses cluster_name instead of hardcoded "Prole CNPG CA"
- registry.py, garage_store.py: refactored into per-mode modules (k3d/k3s/k8s registry and garage store, shared _garage_common)
Deploy / config
- deploy/gcp/gke/knoe-db.yaml: GKE-specific CNPG cluster manifest (rw/ro/r on separate nodes with premium-rwo storage)
- etc/init_common_services.sh, modes/k8s/knoe-db/.version: updated for current deploy
- kong-deployment.yaml: updated manifest
Tests
- test_cluster_nodes_render_smoke.py: add pack/grid, winfo_children, winfo_reqheight, update_idletasks, grid_slaves to dummy widgets; monkeypatch tk.Label so CNPG placement render completes without a real Tkinter root
Co-authored-by: Junie <junie@jetbrains.com>
## GCP / Cluster Environment Screen
- Auto-populate Cloud tab from conf/prod/gcp.cfg on screen open (org_id,
billing_account, billing_project, project_id)
- gcloud auth validity checked on screen startup; friendly modal dialog
streams gcloud auth login output live so user never leaves the app
- Live GKE cluster browser: fetches clusters via gcloud container clusters
list, displays with checkmark selector, auto-selects saved cluster
- Selecting a cluster runs get-credentials, sets KUBECONFIG/KUBECONTEXT,
and syncs the region dropdown to the selected cluster's location
- Region dropdown populated live from gcloud compute regions list with
checkmark on currently selected region; graceful fallback when offline
- New 'GCP Storage' tab with workload->StorageClass mapping (CNPG->premium-rwo,
Redis/Monitoring->standard-rwo, Garage->garage-hdd) and Fetch from Cluster
- Provider readonly field styled correctly (no solid-black on macOS)
- Stale prole.cfg/conf/prole.cfg symlinks removed; all config I/O now
resolves env-specific paths via prole_conf.entrypoint_path()
## GKE Autopilot Compatibility (Common Services)
- Synology iSCSI StorageClass and static PVs guarded behind PROLE_MODE!=k8s
in init_openbao.sh (GKE Autopilot forbids hostPath/iSCSI volumes)
- In-cluster Docker registry (hostPath) skipped in k8s mode; GCP Artifact
Registry used instead
- Kong renamed knoe-svc-kong in k8s mode; all health-check kubectl calls in
init_common_services.sh and status_common_services.sh updated accordingly
- DNS endpoints switched from *.prole.org to *.knoe.dev in k8s mode
(api.knoe.dev, git.knoe.dev, svc.knoe.dev); ingress uses gce class
- New GKE-clean Kong manifests under deploy/opentofu/k8s/manifests/prole/:
no k3s node affinity, explicit Autopilot resource requests/limits
## Garage S3 Store (GKE)
- New garage-statefulset-gcp.yaml targeting garage-hdd StorageClass
(pd-standard, avoids SSD_TOTAL_GB quota exhaustion in us-west3)
- New storageclass-gcp-hdd.yaml (pd-standard, Retain, WaitForFirstConsumer)
- GCP StorageClass manifests skipped on re-runs (Autopilot built-ins are
immutable; skip-if-exists guard added)
- PVC deletion guard extended to cover any storageClass (not just synology)
so stale claims are cleaned before StatefulSet recreation
## Topology (GKE Autopilot)
- DaemonSet collector skipped in prod mode (forbidden in kube-system by
GKE Warden); Kubernetes-only node facts path used instead
- All ready GKE nodes assumed cnpg-eligible and monitoring-eligible without
taint/synology-mount checks (skip_collector + assume_nodes_eligible flags)
## KUBECONFIG / kubectl (k8s mode)
- actions.py: new elif mode==k8s branch sets KUBECONFIG=~/.kube/config
and injects KUBECONTEXT from prole_cfg_data into script env
- _build_kubectl_cmd falls back to Global.KUBECONTEXT when
init_cluster.selected_kubectx is empty
- _activate_selected_gke_cluster persists KUBECONFIG/KUBECONTEXT to
prole_cfg_data and saves prole.cfg immediately after get-credentials
## Database Build Screen (GKE)
- Registry display shows correct Artifact Registry URL
(<region>-docker.pkg.dev/<project>/<namespace>/knoe-db) in green
- Build+push: gcloud auth configure-docker, auto-creates AR repository
named after SERVICE_NAMESPACE (e.g. knoe-system) if missing, then
docker tag + push; falls back to gcr.io if region unavailable
- GCP config loaded from conf/prod/gcp.cfg on every screen entry;
keys normalised to lowercase so project_id lookup is always consistent
## Config / Namespace persistence
- prole_conf.py activate_environment: symlink creation removed; sets
CLUSTER_ENV env-var so all subsequent calls resolve correct env directory
- knoe/ui/screens/__init__.py: startup config load uses entrypoint_path()
instead of hardcoded conf/prole.cfg; seeds SERVICE_NAMESPACE=knoe-system
for managed envs so Common Services never defaults to 'default'
- cfg.py _save_prole_cfg: saves to env-specific path via entrypoint_path()
- etc/prole_cfg.sh: removed all ln -snf symlink creation
Co-authored-by: Junie <junie@jetbrains.com>
- add reusable storage probing subsystem with discovery, bounded probe execution, IO classification, caching, and topology integration
- render per-node storage inventory in Cluster Nodes UI and extend installer test coverage for topology/storage behavior
- introduce core service operation modules and align actions, milestones, services, and supporting configs/scripts for repair/update workflows
- update CNPG/Supabase/database artifacts, placement and port mapping configs, plus related integration tests
Co-authored-by: Junie <junie@jetbrains.com>
All active runtime paths that previously shelled out to etc/init_cloudnative_pg.sh
now dispatch through Python. The shell script has been removed from git.
New files:
- knoe/core/ops/cloudnative_pg.py: canonical Python owner for CNPG lifecycle with
public API (initialize, deploy, rollout, ensure_operator, pin_controller,
install_barman_plugin) and internal helpers (_apply_manifest, _wait_cnpg_pods,
_reconcile_instances, etc.)
- prole/tools/run_cnpg_coverage.py: coverage entry points for k3d/k3s modes,
report sub-command, and check-shell scanner to confirm no live Python dispatch
to the removed shell script
Modified files:
- knoe/core/actions.py: replace shell dispatch in _step_init_scripts,
_step_cnpg_deploy, and repair pipeline with Python calls
- knoe/core/milestones.py: replace shell dispatch in InitializationScriptsMilestone
and DeploymentMilestone
- knoe/ui/screens/services.py: replace shell dispatch for init-scripts step,
deploy button, and rollout button
- prole/deployment.py: replace shell dispatch in _run_post_apply_scripts
- status.py: remove init_cloudnative_pg.sh from _STATUS_SCRIPTS list
- tests/installer/test_actions_helpers.py: mock Python functions, assert shell
script is never dispatched
- tests/installer/test_milestones.py: same
- tests/installer/test_services_init_scripts.py: same
Removed:
- etc/init_cloudnative_pg.sh (git rm)
Verification:
- prole.tools.run_cnpg_coverage check-shell reports clean
- All 61 tests in the affected test files pass
Co-authored-by: Junie <junie@jetbrains.com>
Add storage ops for deterministic namespace+cluster local PV paths, labels/selectors, host-path preparation, and idempotent reconciliation before CNPG cluster apply.
Wire selector injection and validation into CNPG deploy flow/script, and extend installer tests for provisioning orchestration and failure handling.
Co-authored-by: Junie <junie@jetbrains.com>
