Mukul SharmaandClaude Opus 5
7fb35d4062
Node trust covers image pulls, which containerd performs on the node. A docker push from a build pod is a different client with its own trust store and knows nothing about that, so without this a push fails TLS verification while a pull of the same image succeeds — which reads like a broken registry rather than a missing trust anchor. Committed in the clear rather than routed through Vault: this is the CA's public half, the same certificate every node already fetches from Secret Manager. The private key stays in Terraform state and reaches the cluster only as the cert-manager Secret that signs. Consumers mount it at /etc/docker/certs.d/harbor.35.238.248.203.nip.io/ca.crt — the directory name must be the registry hostname exactly, since dockerd looks the path up by host and a mismatch silently does nothing. First consumer is the base-images mirror pod; devops-lib-gcp's build pod follows. Verified the embedded PEM round-trips: its SHA-256 fingerprint matches terraform output registry_ca_cert_pem exactly. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01LEsTefWWifp4ikvhHF5s6N
devops-infra-argo-config
GitOps control plane for infrastructure tooling across Meesho's Kubernetes fleet.
This repo manages ArgoCD Application resources for every infrastructure tool (Contour, VictoriaMetrics, Grafana, Kyverno, KEDA, external-secrets, Vault, etc.) deployed across ~19 clusters. It uses an App-of-Applications pattern: one parent Application per cluster renders child Applications from a appSpec[] list via a generic Helm chart.
Each environment tracks a dedicated branch — merging to that branch triggers immediate ArgoCD auto-sync with no staging gate:
| Environment | Branch |
|---|---|
| Production (prd) | main |
| Staging (stg) | develop |
| Integration (int) | pre-prod |
How it works
incubator/<env>/<cluster>.yaml ← Parent Application (one per cluster)
└── points at generic-argo-apps-chart/ + values/<env>/<cluster>-values.yaml
└── renders one child Application per appSpec[] entry
└── sources charts + overrides from devops-infra-helm-charts
Directory structure
| Directory | Purpose |
|---|---|
incubator/<env>/ |
Parent ArgoCD Application YAML, one per cluster |
values/<env>/ |
Values files defining which tools deploy per cluster |
generic-argo-apps-chart/ |
Helm chart that renders child Applications from appSpec[] |
projects/ |
ArgoCD AppProject definitions (sre, sec) |
external-name-service-*/ |
Cross-cluster DNS routing (ExternalName / MCS topology) |
docs/ |
Agent-facing operational documentation |
skills/ |
Parameterized agent tasks for common operations |
wiki/ |
Architecture decisions and entity pages |
Getting started
- Agents: Read CLAUDE.md first.
- New team members: Read index.md for full navigation.
- PR reviewers: Check docs/global/coding-guidelines/infra-argo.md.
Common operations
| Task | Procedure |
|---|---|
| Add a tool to a cluster | docs/platform/procedures/add-tool-to-cluster.md |
| Upgrade a chart version | docs/platform/procedures/upgrade-chart-version.md |
| Onboard a new cluster | docs/platform/procedures/add-new-cluster.md |
| Roll out a tool fleet-wide | docs/platform/procedures/fleet-wide-tool-rollout.md |
| Debug sync failure | docs/platform/runbooks/argocd-sync-failure.md |
| Debug Helm render error | docs/platform/runbooks/render-failure.md |
| Find values inconsistencies across clusters | docs/platform/runbooks/values-drift.md |
| Debug stuck deployment | docs/platform/runbooks/deployment-stuck.md |
Sister repos
devops-infra-helm-charts— Helm charts andcustom-values.yamloverrides. EveryappSpec[].chartDirandvaluesDirmust exist here.devops-argo-config— Same pattern for service/application workloads (not infra tooling).