Establish clean homelab infrastructure baseline
lint / yaml (push) Has been cancelled
lint / ansible (push) Has been cancelled
lint / terraform (push) Has been cancelled

Reorganize the brownfield repository, remove retired and generated artifacts, harden ignore rules, and record the GitOps/IaC redesign.
This commit is contained in:
2026-09-09 16:47:20 +00:00
commit 88a02ababa
418 changed files with 50579 additions and 0 deletions
+2
View File
@@ -0,0 +1,2 @@
# Real Cloudflare API token — keep secret.example.yaml as the committed template.
secret.yaml
+113
View File
@@ -0,0 +1,113 @@
# cert-manager — certificate issuance
Gives LAN services on `ad.ddupan.top` real, auto-renewing X.509 certs. Added 2026-07-25
so NetBox could be served at `https://netbox.ad.ddupan.top` without a hand-issued cert.
- Chart: `jetstack/cert-manager` **v1.21.0**, namespace `cert-manager`
- Two `ClusterIssuer`s, because the homelab has two ACME sources
| issuer | CA | challenge | use it when |
|---|---|---|---|
| `letsencrypt` | Let's Encrypt (public) | DNS-01 via Cloudflare | browsers must trust it with no CA install |
| `bao-acme` | OpenBao internal PKI (`../../infrastructure/openbao`) | HTTP-01 via the shared gateway | no WAN dependency, or the name must stay out of CT logs |
Both were verified `Ready=True` (ACME account registered) on creation.
## The wildcard, and why
`certificate-wildcard-ad.yaml` issues **one** `*.ad.ddupan.top` cert into
`envoy-gateway-system/wildcard-ad-ddupan-top-tls`, which the shared Gateway's `https`
listener serves for every LAN service (`../envoy-gateway/gateway.yaml`).
Consequences worth understanding:
- **Adding a service costs no certificate work.** An `HTTPRoute` plus an A record in
`samba_ad_extra_a_records` (`../../infrastructure/samba-ad`) is the whole job. No Gateway edit.
- **Internal hostnames stay out of Certificate Transparency logs.** Every Let's Encrypt
issuance is published publicly; per-host certs would make the internal estate
enumerable by anyone. One wildcard entry leaks one name.
- The Secret **must** live in `envoy-gateway-system` (the Gateway's namespace): a listener may only reference a
Secret in the Gateway's own namespace unless a `ReferenceGrant` exists.
- `*.ad.ddupan.top` matches one label only, so the apex `ad.ddupan.top` is listed as an
explicit second SAN.
## Two traps that will cost you an hour each
**1. DNS-01 self-check must not use the cluster resolver.**
cert-manager polls authoritative nameservers for its `_acme-challenge` TXT record before
asking the CA to validate. In-cluster, `ad.ddupan.top` is routed straight to the Samba AD
DC (`../k3s/coredns-custom.yaml`), which is authoritative internally and knows nothing
about a TXT record written into the **public** Cloudflare zone — so the self-check spins
forever while the record is plainly there. Fixed in `values.yaml`:
```yaml
dns01RecursiveNameservers: "1.1.1.1:53,8.8.8.8:53"
dns01RecursiveNameserversOnly: true
```
**2. DNS-01, not HTTP-01, for Let's Encrypt here.** `ad.ddupan.top` resolves only on the
LAN, so LE cannot reach the host to validate. DNS-01 needs no inbound reachability and is
the only challenge that can issue a wildcard.
## Deploy
```bash
helm repo add jetstack https://charts.jetstack.io && helm repo update jetstack
helm upgrade --install cert-manager jetstack/cert-manager --version v1.21.0 \
-n cert-manager --create-namespace -f cert-manager/values.yaml
cp cert-manager/secret.example.yaml cert-manager/secret.yaml # then fill in the token
kubectl apply -f cert-manager/secret.yaml
kubectl apply -f cert-manager/clusterissuer-letsencrypt.yaml \
-f cert-manager/clusterissuer-bao-acme.yaml
kubectl apply -f cert-manager/certificate-wildcard-ad.yaml
```
The Cloudflare token is the same one `cloudflared/terraform` uses. Extract it with:
```bash
sed -nE 's/^[[:space:]]*cloudflare_api_token[[:space:]]*=[[:space:]]*"([^"]*)".*/\1/p' \
cloudflared/terraform/terraform.tfvars
```
> A naive `cut -d= -f2` also swallows the trailing comment on that line and yields a
> 126-character "token" that fails with `Invalid format for Authorization header`.
The Secret must be in the **cert-manager** namespace — a ClusterIssuer resolves solver
Secret refs where cert-manager runs, not where the Certificate lives.
## ⚠ Token scope
That token carries **Account·Cloudflare Tunnel:Edit** as well as Zone·DNS:Edit, so
anything able to read the Secret could rewrite tunnel routing, not just DNS. Reusing it
was a deliberate call to avoid blocking; minting a token scoped to Zone·DNS:Edit on
`ddupan.top` and swapping it in is a clean, self-contained follow-up.
## Verify
```bash
kubectl get clusterissuer # both Ready=True
kubectl -n envoy-gateway-system get certificate # wildcard Ready=True
kubectl -n envoy-gateway-system get order,challenge # empty once issued
kubectl -n envoy-gateway-system get secret wildcard-ad-ddupan-top-tls \
-o jsonpath='{.data.tls\.crt}' | base64 -d | openssl x509 -noout -subject -issuer -enddate
```
Renewal is automatic at 30 days remaining (`renewBefore: 720h`), a deliberately wide
window so a flaky WAN has many chances to retry.
## Using the internal CA instead
Point a Certificate at `bao-acme` when CT-log exposure or WAN dependence matters:
```yaml
issuerRef:
name: bao-acme
kind: ClusterIssuer
```
Issuance is capped by `default_directory_policy = role:bao-server`
(`../../infrastructure/openbao/terraform/pki.tf`), which permits `ad.ddupan.top` subdomains only. Clients
need the internal CA in their trust store — already true for the PVE nodes, the DC and
Authelia, generally **not** for a browser.
@@ -0,0 +1,54 @@
# Cert for the PUBLIC Authelia hostname, so it can also be served on the LAN.
#
# WHY this exists at all, when ../envoy-gateway/gateway.yaml says adding a service
# needs no cert work: that promise holds only for `*.ad.ddupan.top`. Authelia is
# reached at `auth.ddupan.top` — a different zone (Cloudflare is authoritative for
# ddupan.top; the DC is authoritative only for ad.ddupan.top) and one label
# shallower, so the wildcard cannot cover it.
#
# WHY serve a public name internally (split-horizon) rather than introduce an
# internal alias: the OIDC issuer, every registered redirect_uri, and the session
# cookie domain are all `auth.ddupan.top`. Changing the name Gitea talks to would
# mean re-registering every client. Resolving the SAME name to the LAN changes
# nothing Authelia knows about itself.
#
# WHAT BROKE WITHOUT IT (2026-07-28): auth.ddupan.top resolves to Cloudflare proxy
# IPs (104.21.6.55 / 172.67.154.245). TCP/443 to both fails from this network,
# persistently, while other Cloudflare IPs (104.16.132.229) connect fine. Gitea's
# chart runs `gitea admin auth update-oauth` in an INIT container, which fetches
# the discovery URL on every pod start — so Gitea CrashLoopBackOff'd on any
# restart, and server-side token exchange timed out. Routing the name to the LAN
# removes the public internet from an entirely in-cluster conversation. See
# CLAUDE.md: "Internal name resolution must never depend on the WAN."
#
# CT-log note: ../cert-manager/certificate-wildcard-ad.yaml deliberately uses a
# wildcard to keep internal hostnames out of Certificate Transparency logs. That
# reasoning does not apply here — auth.ddupan.top is already public in CT via the
# Cloudflare-facing cert, so naming it costs nothing.
#
# Lives in envoy-gateway-system because a Gateway listener may only reference a
# Secret in the Gateway's own namespace.
---
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
name: auth-ddupan-top
namespace: envoy-gateway-system
spec:
secretName: auth-ddupan-top-tls
issuerRef:
name: letsencrypt
kind: ClusterIssuer
group: cert-manager.io
commonName: "auth.ddupan.top"
dnsNames:
- "auth.ddupan.top"
# DNS-01 via Cloudflare (the letsencrypt ClusterIssuer's solver). Cloudflare stays
# authoritative for ddupan.top, so the challenge resolves publicly even though the
# A record we serve internally points at the LAN.
duration: 2160h # 90d — Let's Encrypt maximum
renewBefore: 720h # 30d
privateKey:
algorithm: ECDSA
size: 256
rotationPolicy: Always
@@ -0,0 +1,44 @@
# Cert for the PUBLIC Gitea hostname, so it can also be served on the LAN.
#
# Same reasoning as certificate-auth-ddupan.yaml: `*.ad.ddupan.top` cannot cover
# `git.ddupan.top` — different zone (Cloudflare is authoritative for ddupan.top,
# the DC only for ad.ddupan.top) and one label shallower.
#
# WHY serve the public name internally rather than introduce git.ad.ddupan.top:
# the remote URL ends up in every clone, every CI checkout, and every existing
# working copy. Split-horizon on the SAME name means none of that has to change,
# and a laptop that leaves the LAN still reaches Gitea through the tunnel with the
# identical URL.
#
# WHAT IT AVOIDS: without this, `git push` goes laptop -> Cloudflare -> tunnel ->
# back into the cluster the laptop is hosting. On 2026-07-28 that path was
# blackholed for hours by a dead VPN tunnel, and the repo is exactly what you need
# during an incident.
#
# CT-log note: the wildcard in certificate-wildcard-ad.yaml exists to keep internal
# hostnames out of Certificate Transparency. That does not apply here —
# git.ddupan.top is already public via the Cloudflare-facing cert.
---
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
name: git-ddupan-top
namespace: envoy-gateway-system
spec:
secretName: git-ddupan-top-tls
issuerRef:
name: letsencrypt
kind: ClusterIssuer
group: cert-manager.io
commonName: "git.ddupan.top"
dnsNames:
- "git.ddupan.top"
# DNS-01 via Cloudflare, which stays authoritative for the zone — so the ACME
# challenge resolves publicly even though the A record we serve on the LAN
# points at the gateway.
duration: 2160h # 90d — Let's Encrypt maximum
renewBefore: 720h # 30d
privateKey:
algorithm: ECDSA
size: 256
rotationPolicy: Always
@@ -0,0 +1,41 @@
# One wildcard cert serving every LAN service on the shared Contour gateway.
#
# WHY A WILDCARD rather than a cert per service:
# 1. Certificate Transparency. Per-host LE certs publish every internal hostname
# to public CT logs, making the whole internal estate enumerable. A single
# *.ad.ddupan.top entry leaks one name and hides the rest.
# 2. The Gateway's HTTPS listener needs exactly one certificateRef to cover all
# hostnames; a new service then needs only an HTTPRoute + a DNS A record, with
# no cert work and no Gateway edit at all.
# 3. Fewer ACME orders against Let's Encrypt rate limits.
#
# Lives in envoy-gateway-system because a Gateway listener may only reference a Secret
# in the Gateway's OWN namespace (cross-namespace refs need a ReferenceGrant).
# Was projectcontour until the gateway moved to Envoy Gateway — see ../envoy-gateway.
#
# NOTE the bare apex `ad.ddupan.top` is listed as well: a wildcard covers
# one label only, so `*.ad.ddupan.top` does NOT match `ad.ddupan.top` itself.
---
apiVersion: cert-manager.io/v1
kind: Certificate
metadata:
name: wildcard-ad-ddupan-top
namespace: envoy-gateway-system
spec:
secretName: wildcard-ad-ddupan-top-tls
issuerRef:
name: letsencrypt
kind: ClusterIssuer
group: cert-manager.io
commonName: "*.ad.ddupan.top"
dnsNames:
- "*.ad.ddupan.top"
- "ad.ddupan.top"
# Renew with 30 days to spare. The WAN is unreliable, so leave a wide window for
# retries rather than the default cutting it fine.
duration: 2160h # 90d — Let's Encrypt maximum
renewBefore: 720h # 30d
privateKey:
algorithm: ECDSA
size: 256
rotationPolicy: Always
@@ -0,0 +1,43 @@
# OpenBao's internal PKI over ACME (../../infrastructure/openbao/terraform/pki.tf).
#
# WHEN TO PREFER THIS OVER letsencrypt:
# * the name must never appear in a public Certificate Transparency log
# * issuance/renewal must not depend on the WAN (see netbox/CONTEXT.md §6)
# * it is a non-web service (LDAPS, Postgres, syslog) where "browser trusts it
# out of the box" buys nothing and the internal CA is already distributed
#
# COST: clients must trust the ddupan.top internal CA. Already true for the PVE
# nodes (pve_ca_trust), Authelia, and the DC — generally NOT true of a fresh browser.
#
# Verified 2026-07-25:
# * directory live, "externalAccountRequired": false -> no EAB stanza needed,
# matching acme_eab_policy = "not-required" in openbao/terraform
# * bao.ad.ddupan.top:8200 serves a REAL Let's Encrypt cert (issuer CN=YE1), so
# cert-manager validates it against public roots — no spec.acme.caBundle required
# * issuance is capped by default_directory_policy = role:bao-server, which permits
# subdomains of ad.ddupan.top only
---
apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
name: bao-acme
spec:
acme:
server: https://bao.ad.ddupan.top:8200/v1/pki/acme/directory
# OpenBao ignores the contact address, but ACME clients must send something.
email: [email protected]
privateKeySecretRef:
name: bao-acme-account-key
solvers:
# http-01, not dns01: bao resolves ad.ddupan.top and can reach LAN hosts
# directly (noted as verified in openbao/terraform/pki.tf), so it can fetch
# the challenge over the LAN with no public exposure. cert-manager creates a
# temporary HTTPRoute on the shared Contour gateway to answer it.
- http01:
gatewayHTTPRoute:
parentRefs:
- name: contour-gateway
namespace: projectcontour
kind: Gateway
group: gateway.networking.k8s.io
sectionName: http # the plaintext :80 listener
@@ -0,0 +1,37 @@
# Let's Encrypt via DNS-01 (Cloudflare).
#
# WHY DNS-01 and not HTTP-01: ad.ddupan.top names resolve ONLY on the LAN, so
# Let's Encrypt cannot reach http://<host>/.well-known/... to validate. DNS-01
# proves control of the name by writing a TXT record into the PUBLIC ddupan.top
# Cloudflare zone, which needs no inbound reachability at all. It is also the only
# challenge type that can issue a WILDCARD.
#
# CT-LOG NOTE: every LE-issued name is published to Certificate Transparency logs.
# Issuing the single wildcard *.ad.ddupan.top (see certificate-wildcard-ad.yaml)
# means only that one entry appears — individual internal hostnames stay private.
# For anything that must not appear at all, use the bao-acme issuer instead.
---
apiVersion: cert-manager.io/v1
kind: ClusterIssuer
metadata:
name: letsencrypt
spec:
acme:
server: https://acme-v02.api.letsencrypt.org/directory
email: [email protected]
# Account key. cert-manager creates this; losing it just means a new account.
privateKeySecretRef:
name: letsencrypt-account-key
solvers:
- dns01:
cloudflare:
# Token is reused from cloudflared/terraform (see cert-manager/README.md).
# ⚠ It also carries Account·Cloudflare Tunnel:Edit, so anything able to
# read this Secret can rewrite tunnel routing, not just DNS. Narrowing it
# to Zone·DNS:Edit is a worthwhile follow-up.
apiTokenSecretRef:
name: cloudflare-api-token
key: api-token
selector:
dnsZones:
- ddupan.top
+20
View File
@@ -0,0 +1,20 @@
# Template for cert-manager/secret.yaml (gitignored). Copy, fill in, apply.
#
# Same token as cloudflared/terraform/terraform.tfvars — extract it with:
# sed -nE 's/^[[:space:]]*cloudflare_api_token[[:space:]]*=[[:space:]]*"([^"]*)".*/\1/p' \
# cloudflared/terraform/terraform.tfvars
# (a naive `cut -d= -f2` also swallows the trailing comment on that line)
#
# Must live in the cert-manager namespace: a ClusterIssuer resolves solver Secret
# refs in the namespace where cert-manager runs, not where the Certificate is.
#
# ⚠ This token also carries Account·Cloudflare Tunnel:Edit. Replacing it with one
# scoped to Zone·DNS:Edit on ddupan.top would shrink the blast radius to DNS.
apiVersion: v1
kind: Secret
metadata:
name: cloudflare-api-token
namespace: cert-manager
type: Opaque
stringData:
api-token: "REPLACE_WITH_CLOUDFLARE_API_TOKEN"
+55
View File
@@ -0,0 +1,55 @@
# cert-manager — X.509 issuance for the cluster.
#
# Chart: jetstack/cert-manager v1.21.0 (repo: https://charts.jetstack.io)
#
# Exists so LAN services on ad.ddupan.top get real, auto-renewing certs instead of
# hand-issued ones. The homelab has TWO ACME sources and both are wired up as
# ClusterIssuers (see clusterissuer-*.yaml):
# letsencrypt — public CA, DNS-01 via Cloudflare. Browser-trusted with no CA
# install. Used for the *.ad.ddupan.top wildcard.
# bao-acme — OpenBao's internal PKI (../../infrastructure/openbao). No WAN dependency and
# nothing published to Certificate Transparency logs.
# CRDs are part of the release so `helm uninstall` is a clean removal and there is
# no separate kubectl-apply step to forget.
crds:
enabled: true
keep: true # don't let an accidental uninstall garbage-collect live Certificates
# Single-node k3s: one of everything, modest requests. The laptop runs the whole
# homelab (see netbox/CONTEXT.md §6).
replicaCount: 1
resources:
requests:
cpu: 10m
memory: 64Mi
limits:
memory: 192Mi
webhook:
replicaCount: 1
resources:
requests:
cpu: 10m
memory: 32Mi
limits:
memory: 128Mi
cainjector:
replicaCount: 1
resources:
requests:
cpu: 10m
memory: 96Mi
limits:
memory: 256Mi
# ⚠ DNS-01 self-check: cert-manager polls authoritative NS for the _acme-challenge
# TXT record before telling the CA to validate. By default it asks the cluster's
# resolver, which for ad.ddupan.top is CoreDNS -> the Samba AD DC (k3s/coredns-custom.yaml).
# The DC is authoritative for ad.ddupan.top internally and knows nothing about the
# TXT record we just wrote into the PUBLIC Cloudflare zone, so the self-check would
# spin forever. Forcing public recursive resolvers makes the check see what the CA
# will see.
dns01RecursiveNameservers: "1.1.1.1:53,8.8.8.8:53"
dns01RecursiveNameserversOnly: true
+118
View File
@@ -0,0 +1,118 @@
# Envoy Gateway — LAN ingress + Authelia forward-auth
The cluster's HTTP entry point for services on `ad.ddupan.top`, and the enforcement
point for Authelia authentication. Envoy's LoadBalancer holds **192.168.10.127** (k3s
ServiceLB); routing is by `Host` header.
- Chart: `oci://docker.io/envoyproxy/gateway-helm` **v1.5.6**, ns `envoy-gateway-system`
- `gateway.yaml` — `GatewayClass eg` + `Gateway eg` (`:80` plaintext, `:443` wildcard TLS)
- Gateway API **v1.3.0**
- Replaced Contour on 2026-07-25 (see [Why not Contour](#why-not-contour))
## Adding a service
1. an `HTTPRoute` with `parentRefs` → `eg` / `envoy-gateway-system`, `sectionName: https`
2. an A record in `samba_ad_extra_a_records` → `192.168.10.127` (`../../infrastructure/samba-ad`), then
`ansible-playbook provision-dc.yml --tags dns`
3. **optionally** a `SecurityPolicy` for Authelia forward-auth — see `../../apps/netbox`
No certificate work: the `:443` listener already serves the `*.ad.ddupan.top` wildcard
from `../cert-manager`. Worked example: `../../apps/netbox`.
## Why not Contour
Contour worked as an ingress, but supports **only the gRPC** Envoy ext_authz protocol —
*"Only the Envoy GRPC authorization protocol will be supported"* (Contour 1.33 docs).
Authelia implements the **HTTP** ExtAuthz filter. The two cannot meet, so Authelia
forward-auth was impossible.
That mattered because **NetBox has no SSO group→role mapping** — its group/superuser
mapping is LDAP-only, and the OIDC pipeline can only assign one static group. Without
forward-auth the options were "promote every user by hand" or "drop SSO and use LDAP,
losing 2FA". Envoy Gateway's `SecurityPolicy.extAuth.http` removes the dilemma.
Envoy Gateway was chosen over Traefik because everything already built — Gateway,
HTTPRoute, wildcard cert, DNS — is Gateway API, so only the GatewayClass and controller
changed. Traefik would have meant reverting to `IngressRoute` + `Middleware`.
## Forward-auth: how it fits together
```
browser ──▶ Envoy ──(ext_authz HTTP)──▶ Authelia ──▶ 200 + Remote-* headers
│ └──▶ 401 ⇒ 302 to auth.ddupan.top
└──▶ upstream app (headers attached)
```
Pieces, each in the directory that owns it:
| where | what |
|---|---|
| `../../apps/netbox/securitypolicy.yaml` | `SecurityPolicy` targeting the app's HTTPRoute |
| `../../apps/authelia/referencegrant-extauth.yaml` | lets a SecurityPolicy in another namespace reference the Authelia Service |
| `../../apps/authelia/values.yaml` | `server.endpoints.authz.ext-authz` + an `access_control` rule |
| `../../apps/netbox/networkpolicy.yaml` | stops anything bypassing Envoy to reach the app directly |
### Three details that cost time
- **`headersToBackend` lives under `extAuth.http`, not `extAuth`.** One level up the API
rejects it: `unknown field "spec.extAuth.headersToBackend"`.
- **`backendRefs.port` is the SERVICE port, not the container port.** The Authelia chart
publishes `80 → targetPort http (9091)`; using `9091` fails with
`TCP Port 9091 not found on service authelia/authelia`.
- **Authelia's session cookie is scoped to `ddupan.top`**, which already covers
`*.ad.ddupan.top` — so SSO works across both without touching the cookie config.
### ⚠ Trust boundary
Apps behind forward-auth trust `Remote-*` headers. Two things make that safe and **both**
must remain true:
1. `headersToBackend` **overrides** any client-supplied value ("coexisting headers will be
overridden"), so a spoofed `Remote-User` cannot survive the hop through Envoy.
2. A `NetworkPolicy` per app restricts pod ingress to `envoy-gateway-system`, so nothing
in-cluster can bypass Envoy. k3s enforces NetworkPolicy (kube-router), so this is real.
Verified by test: a pod in `default` sending `Remote-User: admin` to the app Service gets
**connection refused**, while the same request from `envoy-gateway-system` is served.
Also set `failOpen: false` — if Authelia is down, refuse traffic rather than admit
unauthenticated requests to an app whose auth model is "trust the header".
## Install
```bash
# The Gateway API CRDs may already be owned by another tool's field manager (Contour's
# quickstart used client-side apply), which makes Helm fail with
# "conflict with kubectl-client-side-apply: .spec.versions".
# Transfer ownership WITHOUT deleting the CRDs (deleting them would delete every
# Gateway and HTTPRoute):
helm pull oci://docker.io/envoyproxy/gateway-helm --version v1.5.6 --untar
kubectl apply --server-side --force-conflicts --field-manager=helm \
-f gateway-helm/crds/gatewayapi-crds.yaml
helm upgrade --install envoy-gateway oci://docker.io/envoyproxy/gateway-helm \
--version v1.5.6 -n envoy-gateway-system --create-namespace
kubectl apply -f envoy-gateway/gateway.yaml
```
Only one LoadBalancer can hold `:80/:443` — k3s ServiceLB uses hostPorts, so a second
one's `svclb-*` DaemonSet sits at `0/1` and the Gateway stays `PROGRAMMED: False` until
the previous controller's Service is gone.
## Verify
```bash
kubectl -n envoy-gateway-system get gateway eg # PROGRAMMED=True, ADDRESS 192.168.10.127
kubectl -n envoy-gateway-system get pods # envoy-... 2/2
kubectl -n <ns> get securitypolicy # Accepted=True
curl -s -o /dev/null -w '%{http_code}\n' http://192.168.10.127/ # 404 = serving, no route matched
```
`ss -lntp` shows **nothing** on `:80/:443` even when healthy — ServiceLB forwards via
iptables/NodePort rather than binding. Test with `curl`, not `ss`.
## Outstanding
Contour source manifests were removed from the clean baseline. Live leftovers may
still exist and must be inventoried before deletion; do not infer live state from
the archive cleanup. The legacy Git history retains the retired manifests.
+116
View File
@@ -0,0 +1,116 @@
# LAN ingress for k3s services on ad.ddupan.top.
#
# Replaced Contour 2026-07-25. Contour was dropped for one concrete reason: it
# supports ONLY the gRPC Envoy ext_authz protocol ("Only the Envoy GRPC authorization
# protocol will be supported" — Contour 1.33 docs), while Authelia implements the
# HTTP ExtAuthz filter. That made Authelia forward-auth impossible, which in turn made
# AD-group -> role mapping impossible for apps like NetBox whose SSO support has no
# group mapping. Envoy Gateway's SecurityPolicy supports extAuth.http, so it works.
#
# Everything else carried over unchanged: the Gateway API objects, the wildcard cert
# from ../cert-manager, and the DNS A records in ../../infrastructure/samba-ad.
---
apiVersion: gateway.networking.k8s.io/v1
kind: GatewayClass
metadata:
name: eg
spec:
controllerName: gateway.envoyproxy.io/gatewayclass-controller
---
apiVersion: gateway.networking.k8s.io/v1
kind: Gateway
metadata:
name: eg
namespace: envoy-gateway-system
spec:
gatewayClassName: eg
listeners:
# Plaintext :80 — ACME http-01 (the bao-acme ClusterIssuer solves through here)
# and HTTP->HTTPS redirects. Never serve anything sensitive on it.
- name: http
protocol: HTTP
port: 80
allowedRoutes:
namespaces:
from: All
kinds:
- group: gateway.networking.k8s.io
kind: HTTPRoute
# HTTPS :443. One wildcard covers every LAN service, so adding a service is an
# HTTPRoute + a DNS A record — no cert work and no edit here.
#
# ⚠ The cert Secret MUST live in this Gateway's namespace (a listener may only
# reference a Secret in its own namespace without a ReferenceGrant), which is why
# cert-manager/certificate-wildcard-ad.yaml issues into envoy-gateway-system.
#
# ⚠ Never point this at a Secret the gateway controller generates for its own
# internal use. Doing exactly that with Contour's `contourcert-*` xDS secret killed
# its dataplane for 141 days — see the legacy Git history.
- name: https
protocol: HTTPS
port: 443
hostname: "*.ad.ddupan.top"
tls:
mode: Terminate
certificateRefs:
- kind: Secret
name: wildcard-ad-ddupan-top-tls
allowedRoutes:
namespaces:
from: All
kinds:
- group: gateway.networking.k8s.io
kind: HTTPRoute
# HTTPS :443 for auth.ddupan.top specifically.
#
# WHY a second listener rather than another hostname on the one above: a
# listener carries exactly one hostname, and `*.ad.ddupan.top` cannot match
# `auth.ddupan.top` — different zone (Cloudflare is authoritative for
# ddupan.top, the DC only for ad.ddupan.top) and one label shallower. Envoy
# selects between them by SNI, so both coexist on :443 cleanly.
#
# This exists so in-cluster clients reach Authelia over the LAN instead of
# hairpinning through Cloudflare proxy IPs that are unroutable from this
# network. Full reasoning in ../cert-manager/certificate-auth-ddupan.yaml;
# the route is ../../apps/authelia/httproute.yaml.
- name: https-auth
protocol: HTTPS
port: 443
hostname: "auth.ddupan.top"
tls:
mode: Terminate
certificateRefs:
- kind: Secret
name: auth-ddupan-top-tls
allowedRoutes:
namespaces:
from: All
kinds:
- group: gateway.networking.k8s.io
kind: HTTPRoute
# HTTPS :443 for git.ddupan.top. Third listener for the same reason as the
# second: one hostname per listener, and the *.ad.ddupan.top wildcard cannot
# match a name in the ddupan.top zone. Envoy picks between all three by SNI.
#
# Exists so `git push` stays on the LAN instead of going out to Cloudflare and
# back down the tunnel into this same cluster. The repo is what you need during
# an incident, so it must not depend on the WAN. Cert:
# ../cert-manager/certificate-git-ddupan.yaml; route: ../../apps/gitea/httproute.yaml.
- name: https-git
protocol: HTTPS
port: 443
hostname: "git.ddupan.top"
tls:
mode: Terminate
certificateRefs:
- kind: Secret
name: git-ddupan-top-tls
allowedRoutes:
namespaces:
from: All
kinds:
- group: gateway.networking.k8s.io
kind: HTTPRoute
@@ -0,0 +1,43 @@
# The one store every namespace reads from.
#
# Cluster-scoped on purpose: authelia, gitea and cloudflared all consume it, and a
# per-namespace SecretStore would mean duplicating the OpenBao connection details
# three times.
#
# AUTH: no credential is stored anywhere. ESO presents its own ServiceAccount JWT,
# OpenBao validates it against the cluster's TokenReview API, and hands back a
# short-lived token scoped by the `external-secrets` role. The reviewer JWT that
# makes that possible lives on the bao host, configured by
# ../../infrastructure/openbao/ansible/roles/openbao_bootstrap/tasks/auth_kubernetes.yml — it is key
# material, which is why Ansible owns it and Terraform does not.
#
# TLS: bao presents a Let's Encrypt cert for bao.ad.ddupan.top, so no caBundle or
# caProvider is needed — verified from inside a pod (HTTP 200, ssl_verify_result 0).
# ⚠ Address it by HOSTNAME, never 192.168.10.8: the cert carries a DNS SAN only,
# so connecting by IP fails verification. Same trap as dc1's LDAPS cert.
---
apiVersion: external-secrets.io/v1
kind: ClusterSecretStore
metadata:
name: openbao
spec:
provider:
vault:
# OpenBao is Vault-API compatible; ESO's vault provider drives it unchanged.
server: 'https://bao.ad.ddupan.top:8200'
# Mount path of the KV engine, from ../../infrastructure/openbao/terraform/mounts.tf.
path: 'kv'
version: 'v2'
auth:
kubernetes:
mountPath: 'kubernetes'
role: 'external-secrets'
serviceAccountRef:
name: 'external-secrets'
# namespace is MANDATORY on a ClusterSecretStore (it has no namespace
# of its own to resolve the reference against).
namespace: 'external-secrets'
# NOTE: Vault 1.21+ requires an `audiences: ['vault']` entry here, and
# the bao role must declare a matching audience. OpenBao 2.6.1 does not,
# so it is omitted — if auth ever starts failing with an audience
# mismatch after an upgrade, this is the first thing to add.
@@ -0,0 +1,119 @@
# One ExternalSecret per Kubernetes Secret that was previously created by hand.
#
# These REPLACE the hand-applied secret.yaml files in ../../apps/authelia, ../../apps/gitea and
# ../../infrastructure/cloudflared. Those files stay gitignored as the break-glass copy — if OpenBao
# is down during a rebuild you can still `kubectl apply -f <svc>/secret.yaml` and
# carry on. ESO will reconcile back over it once bao returns.
#
# ⚠ ESO takes OWNERSHIP of the target Secret. The existing hand-created Secrets have
# the same names on purpose, so ESO adopts rather than duplicates them — but that
# also means a wrong key name here will OVERWRITE a working Secret with an empty or
# partial one. Check `kubectl get externalsecret -A` reports SecretSynced before
# trusting it, and remember Authelia and Gitea only read their secrets at startup.
#
# refreshInterval is deliberately long. Nothing here rotates on its own yet, and a
# short interval only adds load plus a wider window to notice a bad sync. When the
# OpenBao PostgreSQL secrets engine lands (see ../docs/cicd.md) the rotating
# credentials will want a shorter interval AND something to restart the consumer.
---
apiVersion: external-secrets.io/v1
kind: ExternalSecret
metadata:
name: authelia-secrets
namespace: authelia
spec:
refreshInterval: 1h
secretStoreRef:
name: openbao
kind: ClusterSecretStore
target:
name: authelia-secrets
creationPolicy: Owner
# extract pulls EVERY key at the path, so the six key names live in OpenBao
# rather than being restated here. They must match the `path:` values in
# ../../apps/authelia/values.yaml exactly.
dataFrom:
- extract:
key: k8s/authelia
---
# Separate Secret, not a seventh key in authelia-secrets — the chart projects the
# existingSecret volume with an explicit items: list of only the six keys it
# generates, so an extra key would be stored but never mounted. That mistake took
# SSO down on 2026-07-28.
apiVersion: external-secrets.io/v1
kind: ExternalSecret
metadata:
name: authelia-oidc-jwks
namespace: authelia
spec:
refreshInterval: 1h
secretStoreRef:
name: openbao
kind: ClusterSecretStore
target:
name: authelia-oidc-jwks
creationPolicy: Owner
dataFrom:
- extract:
key: k8s/authelia-oidc-jwks
---
apiVersion: external-secrets.io/v1
kind: ExternalSecret
metadata:
name: gitea-db
namespace: gitea
spec:
refreshInterval: 1h
secretStoreRef:
name: openbao
kind: ClusterSecretStore
target:
name: gitea-db
creationPolicy: Owner
dataFrom:
- extract:
key: k8s/gitea
---
apiVersion: external-secrets.io/v1
kind: ExternalSecret
metadata:
name: cloudflared-tunnel
namespace: cloudflared
spec:
refreshInterval: 1h
secretStoreRef:
name: openbao
kind: ClusterSecretStore
target:
name: cloudflared-tunnel
creationPolicy: Owner
dataFrom:
- extract:
key: k8s/cloudflared
---
# SeaweedFS S3 identities, including the least-privilege `terraform` identity that
# holds Terraform remote state.
#
# ⚠ These credentials were previously INLINE in ../../apps/seaweedfs/values.yaml and went
# into git with the initial commit. Externalising stops the bleeding; the leaked
# anvAdmin key still needs rotating separately, since it is in history.
#
# The chart normally GENERATES seaweedfs-s3-secret from s3.credentials. We point
# filer.s3.existingConfigSecret at this one instead, so the chart stops rendering
# credentials from values entirely.
apiVersion: external-secrets.io/v1
kind: ExternalSecret
metadata:
name: seaweedfs-s3-config
namespace: seaweedfs
spec:
refreshInterval: 1h
secretStoreRef:
name: openbao
kind: ClusterSecretStore
target:
name: seaweedfs-s3-config
creationPolicy: Owner
dataFrom:
- extract:
key: k8s/seaweedfs-s3
+43
View File
@@ -0,0 +1,43 @@
# External Secrets Operator — pulls secret material from OpenBao into Kubernetes
# Secrets, so the Secrets themselves become declarative instead of hand-created.
#
# Install:
# helm upgrade --install external-secrets external-secrets/external-secrets \
# -n external-secrets --create-namespace -f values.yaml
#
# WHY this and not SOPS: OpenBao is already the secrets store and the internal CA
# here, and its Kubernetes auth backend is already bootstrapped
# (../../infrastructure/openbao/ansible/roles/openbao_bootstrap/tasks/auth_kubernetes.yml), so a pod
# authenticates with its own ServiceAccount JWT and NOTHING long-lived is stored
# in the cluster. SOPS would mean managing an age key and committing ciphertext.
#
# Single node, so one replica of each component. The webhook and cert-controller
# are not optional — the CRDs use conversion/validating webhooks.
replicaCount: 1
webhook:
replicaCount: 1
resources:
requests: {cpu: 10m, memory: 32Mi}
limits: {memory: 128Mi}
certController:
replicaCount: 1
resources:
requests: {cpu: 10m, memory: 32Mi}
limits: {memory: 128Mi}
resources:
requests: {cpu: 10m, memory: 64Mi}
limits: {memory: 256Mi}
# The controller's own ServiceAccount is what the ClusterSecretStore presents to
# OpenBao, so its name is part of the contract with the bao Kubernetes auth role
# (bound_service_account_names). Pinned rather than left to the chart's default.
serviceAccount:
create: true
name: external-secrets
# ClusterSecretStore is cluster-scoped; leaving this on lets one store serve every
# namespace, which is the point here (authelia, gitea and cloudflared all consume it).
installCRDs: true
+24
View File
@@ -0,0 +1,24 @@
.:53 {
errors
health
ready
kubernetes cluster.local in-addr.arpa ip6.arpa {
pods insecure
fallthrough in-addr.arpa ip6.arpa
}
hosts /etc/coredns/NodeHosts {
ttl 60
reload 15s
fallthrough
}
prometheus :9153
cache 30 {
serve_stale 1h immediate
}
loop
reload
loadbalance
import /etc/coredns/custom/*.override
forward . /etc/resolv.conf
}
import /etc/coredns/custom/*.server
+128
View File
@@ -0,0 +1,128 @@
# Route internal AD-zone lookups straight to the Samba DC instead of the LAN
# router, for every pod in the cluster.
#
# WHY: k3s CoreDNS forwards to the node's /etc/resolv.conf, which lists the
# router (192.168.10.1) — and that resolver flaps. When it hangs, CoreDNS's
# forward plugin blocks and even CLUSTER-INTERNAL lookups
# (*.svc.cluster.local) start timing out, which is how Authelia ended up in
# CrashLoopBackOff: its startup check resolves both dc1.ad.ddupan.top and
# smtp-relay.smtp-relay.svc.cluster.local, and a strict startup check turns a
# transient DNS blip into a restart loop.
#
# dc1 is AUTHORITATIVE for ad.ddupan.top, so sending that zone directly to it
# removes the router from the path entirely — no forwarding, no upstream
# dependency, no flap.
#
# k3s picks this up via `import /etc/coredns/custom/*.server` in its Corefile.
# The `reload` plugin applies it without restarting CoreDNS.
apiVersion: v1
kind: ConfigMap
metadata:
name: coredns-custom
namespace: kube-system
data:
ad-ddupan-top.server: |
ad.ddupan.top:53 {
errors
cache 30
forward . 192.168.10.5 {
policy sequential
}
}
# ── stop search-domain permutations from hanging on the router ──────────
# Pods get `options ndots:5` plus the NODE's search list, which still carries
# `lab.ddupan.top` (the RETIRED domain) and tailscale's MagicDNS suffix. A name
# like dc1.ad.ddupan.top has only 3 dots, so glibc tries EVERY suffix first:
# dc1.ad.ddupan.top.tail7e769.ts.net
# dc1.ad.ddupan.top.lab.ddupan.top
# Both are forwarded upstream to the flaky router and time out, so the client
# gives up before the bare name is ever tried. That is what put Authelia in
# CrashLoopBackOff -- and it hit the cluster-internal smtp-relay name too
# (4 dots < ndots:5).
#
# Answer these locally with an instant NXDOMAIN. Neither zone should ever
# resolve from inside the cluster.
dead-search-domains.server: |
lab.ddupan.top:53 {
errors
template ANY ANY {
rcode NXDOMAIN
}
}
tail7e769.ts.net:53 {
errors
template ANY ANY {
rcode NXDOMAIN
}
}
# ── split-horizon for the Authelia hostname ─────────────────────────────
# Cloudflare is authoritative for ddupan.top (the DC only holds ad.ddupan.top),
# so auth.ddupan.top publicly resolves to Cloudflare proxy IPs — 104.21.6.55 and
# 172.67.154.245. TCP/443 to BOTH fails from this network, persistently, while
# other Cloudflare IPs (104.16.132.229) connect fine. So every in-cluster
# consumer of Authelia was hairpinning out to an internet path that does not
# work, to reach a Service sitting in the same cluster.
#
# Concretely (2026-07-28): the Gitea chart runs `gitea admin auth update-oauth`
# in an INIT container, which FETCHES the OIDC discovery URL on every pod start.
# It timed out, so Gitea CrashLoopBackOff'd on any restart, and server-side
# token exchange failed for logins. This was latent — any restart would do it.
#
# Answer with the Envoy Gateway LAN address instead. The gateway terminates TLS
# with a real Let's Encrypt cert for this exact name
# (../cert-manager/certificate-auth-ddupan.yaml) and routes to the authelia
# Service (../../apps/authelia/httproute.yaml), so the name, issuer, redirect URIs and
# cookie domain are all unchanged — only the path stops leaving the LAN.
#
# A `template` block (not `hosts`) so it answers only A/AAAA-shaped queries and
# returns NOERROR/no-data rather than NXDOMAIN for anything else.
auth-ddupan-top.server: |
auth.ddupan.top:53 {
errors
template IN A {
answer "{{ .Name }} 60 IN A 192.168.10.127"
}
template IN AAAA {
rcode NOERROR
}
}
# ── split-horizon for Gitea ─────────────────────────────────────────────
# Same shape and same reason as auth.ddupan.top above: git.ddupan.top publicly
# resolves to Cloudflare, so an in-cluster client cloning from Gitea would leave
# the LAN, cross the WAN and come back down the tunnel into this same cluster.
#
# This matters for CI: an Actions runner checking out the repo, and anything
# else that clones from inside the cluster. Served by the `https-git` listener
# with a real LE cert for the name (../cert-manager/certificate-git-ddupan.yaml),
# so the clone URL is identical inside and outside — no remote needs rewriting.
#
# ⚠ This only covers PODS. LAN clients resolve via the DC, which forwards
# ddupan.top to the router — and the NEC IX has no static-host/proxy-DNS
# feature (`show dns` offers only fqdn-database). The DC therefore needs its own
# `git.ddupan.top` zone for laptops and PVE nodes to get the LAN answer.
git-ddupan-top.server: |
git.ddupan.top:53 {
errors
template IN A {
answer "{{ .Name }} 60 IN A 192.168.10.127"
}
template IN AAAA {
rcode NOERROR
}
}
---
# NOTE: `serve_stale` could NOT be added here.
#
# The cache plugin already exists in k3s's own Corefile (`cache 30`), and CoreDNS
# rejects a duplicate plugin in the same server block — so the *.override import
# cannot carry it. It was applied by patching the `coredns` ConfigMap directly;
# the intended Corefile is kept alongside as `Corefile.desired`.
#
# ⚠️ A k3s restart/upgrade re-applies its bundled manifest and will REVERT that
# patch. Re-apply from Corefile.desired if external name resolution starts
# failing hard during WAN outages again.
#
# WHY: the ISP drops the WAN at random and it cannot be changed. serve_stale
# keeps answering with expired entries while the upstream is unreachable, so
# previously-resolved external names (smtp.office365.com, package mirrors) keep
# working through a blip instead of hanging.
+138
View File
@@ -0,0 +1,138 @@
# observability
Metrics, logs, and traces for the cluster — one VictoriaMetrics-ecosystem stack,
operator-driven, running in the `monitoring` namespace. Replaces the standalone
Docker Compose stack in `../../apps/victoriametrics/`.
## Status — DEPLOYED (2026-07-10)
Live and verified in the `monitoring` namespace (operator chart 0.66.2):
metrics (data queryable), logs (pods ingesting), traces (VTSingle CRD, verified via
Jaeger API), Grafana (3 datasources healthy, VM dashboard loaded, Tailscale ingress
`grafana.tail7e769.ts.net`, Authelia OIDC client registered). All PVCs on
`localpv-zfs-ceph`. Note the vmagent hot-reload RBAC fix (`metrics/reload-rbac.yaml`,
see `metrics/README-reload.md`) and that VL/VT CRDs are served under `.../v1`.
Still host-side (not yet done): deploy `docker-hosts/compose.yaml` on the Docker
hosts and `logs/vlogs-ingress.yaml` to push their logs.
## Decision
| Pillar | Choice | Why |
| -------- | ----------------------------- | --- |
| Metrics | **VictoriaMetrics** (VMSingle)| already in use; keep it |
| Logs | **VictoriaLogs** (VLSingle) | same vendor/operator, tiny footprint, LogsQL |
| Traces | **VictoriaTraces** + OTel Collector | no extra storage deps, OTLP in, Jaeger query API |
| Collect | vmagent (metrics), vlagent (logs), OTel Collector (traces) | native to each backend |
| Visualize| **Grafana**, one instance, 3 datasources | single pane |
| Manage | **victoria-metrics-operator** | VMSingle/VMAgent/VMAlert/VMAlertmanager/VMRule **and** VLSingle as CRDs |
| Expose | **Tailscale ingress** (private) + **Authelia OIDC** | admin tool: private + SSO |
Grafana's Prometheus-operator converter is on, so any chart shipping a
`ServiceMonitor`/`PodMonitor`/`PrometheusRule` is scraped automatically.
## Architecture
```
Grafana (grafana.<tailnet>.ts.net, Authelia OIDC)
├── VictoriaMetrics datasource → VMSingle ← vmagent (k8s SD, node-exporter, kubelet/cAdvisor, KSM)
├── VictoriaLogs datasource → VLSingle ← vlagent DaemonSet (pod logs)
└── Jaeger datasource → VictoriaTraces ← OTel Collector (OTLP 4317/4318) ← apps
VMAlert (VMRules) → VMAlertmanager
```
## Layout
```
observability/
namespace.yaml monitoring namespace
operator/ victoria-metrics-operator (Helm)
metrics/
vmsingle|vmagent|vmalert|vmalertmanager.yaml core CRs
rules/ VMRules ported from ../../apps/victoriametrics/rules
exporters/ node-exporter (+VMNodeScrape), kube-state-metrics (Helm)
scrapes/ kubelet + cAdvisor VMNodeScrapes
logs/ VLSingle CR + victoria-logs-collector (Helm)
traces/ victoria-traces-single (Helm) + OTel Collector
grafana/ Grafana (Helm) values, OIDC secret, install script
docker-hosts/ per-Docker-host sidecar (cAdvisor + node-exporter + Vector)
```
## Docker hosts
The Compose services on Docker hosts (`../../apps/vlmcsd`, `../../apps/ps3netsrv`, `../../apps/netboot`, ...
and the legacy `../../apps/victoriametrics` stack) are monitored too:
- **Metrics — pull, nothing exposed.** Each host runs `docker-hosts/compose.yaml`
(cAdvisor `:8080` + node-exporter `:9100`). The cluster's vmagent scrapes their LAN
IPs via `metrics/scrapes/docker-hosts.yaml` (VMStaticScrape). VMSingle stays private.
- **Logs — push over the tailnet.** A **Vector** container per host tails the Docker
socket and ships to VictoriaLogs' Elasticsearch-bulk endpoint, exposed privately as
`vlogs.<tailnet>.ts.net` by `logs/vlogs-ingress.yaml`.
Deploy:
```bash
# cluster side (once): expose VictoriaLogs ingest + register the scrape
kubectl apply -f logs/vlogs-ingress.yaml -f metrics/scrapes/docker-hosts.yaml
# on each Docker host:
cd docker-hosts
cp .env.example .env # set HOST_LABEL and confirm VLOGS_ENDPOINT (tailnet FQDN)
docker compose up -d
# then add the host's IP to metrics/scrapes/docker-hosts.yaml and re-apply
```
Note: cAdvisor uses host port `8080` — remap it (and the scrape target) if taken.
## Deploy order
```bash
# 1. Operator (installs CRDs) + namespace
cd operator && ./helm.sh && cd ..
# 2. Metrics: core CRs, rules, scrapes, exporters
kubectl apply -f metrics/vmsingle.yaml -f metrics/vmagent.yaml \
-f metrics/vmalert.yaml -f metrics/vmalertmanager.yaml
kubectl apply -f metrics/rules/ -f metrics/scrapes/
kubectl apply -f metrics/exporters/node-exporter.yaml
metrics/exporters/kube-state-metrics.sh # needs Prometheus CRDs; see the script
# 3. Logs
cd logs && ./helm.sh && cd ..
# 4. Traces
cd traces && ./helm.sh && cd ..
# 5. Grafana (create the OIDC secret first — see grafana/oidc-secret.yaml)
cd grafana && ./helm.sh && cd ..
```
## Before it works — required edits
- **OIDC secret pair.** Generate once:
`authelia crypto hash generate pbkdf2 --variant sha512 --random --random.length 72`
- plaintext → the `grafana-oidc` Secret (create imperatively, don't commit)
- hash → `authelia/values.yaml` grafana client `client_secret`, then `helm upgrade` Authelia.
- **Tailnet FQDN.** This scaffold assumes `grafana.tail7e769.ts.net` (your existing tailnet,
per `seaweedfs-admin`). If the operator assigns a different name, update it in
`grafana/values.yaml` (`root_url`) **and** the Authelia grafana `redirect_uris`.
- **Chart value names.** `logs/collector-values.yaml` and `traces/traces-values.yaml` note a
`helm show values` check — verify the keys against the installed chart versions.
- **Pin chart + image versions** after first install (all `helm.sh` scripts note this;
`otel-collector.yaml` pins the collector image — bump to current).
## Follow-ups
- Dashboards: the VictoriaMetrics single-node board is migrated to
`grafana/dashboards/victoriametrics.json`; `grafana/dashboards/apply.sh` loads it via
the Grafana sidecar. (The old `vmagent.json`/`vmalert.json` were empty dirs Docker
auto-created, not real boards.) Add the canonical ones by dropping their JSON in
`grafana/dashboards/` and re-running the script — vmagent (gnetId 12683),
vmalert (14950), node-exporter (1860), cAdvisor now that those exporters exist.
- Wire real Alertmanager receivers in `metrics/vmalertmanager.yaml` (email via
`../../apps/smtp-relay/`), replacing the ported `blackhole`.
- Once parity is confirmed, decommission the Compose stack: `../../apps/victoriametrics/`
(`docker compose down`), and retire that folder.
- Add app instrumentation: point `OTEL_EXPORTER_OTLP_ENDPOINT` at
`otel-collector.monitoring.svc:4317`.
@@ -0,0 +1,9 @@
# Copy to .env and adjust per host. Do not commit the real .env.
# A stable name for THIS Docker host; becomes the `host` log stream label and the
# `host` label on its scraped metrics (keep it consistent with the VMStaticScrape).
HOST_LABEL=docker-01
# VictoriaLogs Elasticsearch-bulk ingest endpoint, reached over the tailnet
# (exposed by ../logs/vlogs-ingress.yaml). Confirm the tailnet FQDN after applying.
VLOGS_ENDPOINT=https://vlogs.tail7e769.ts.net/insert/elasticsearch/
@@ -0,0 +1,50 @@
# Per-Docker-host monitoring sidecar. Deploy this compose on every Docker host you
# want visible in the observability stack. It exposes host + container metrics for the
# cluster's vmagent to PULL over the LAN, and pushes container logs to VictoriaLogs.
#
# cp .env.example .env # set HOST_LABEL + VLOGS_ENDPOINT
# docker compose up -d
#
# Then add this host's IP to ../metrics/scrapes/docker-hosts.yaml (VMStaticScrape).
services:
# Host metrics (CPU/mem/disk/net/fs). host networking -> listens on <hostIP>:9100.
node-exporter:
image: quay.io/prometheus/node-exporter:v1.9.1
command:
- --path.rootfs=/host
network_mode: host
pid: host
volumes:
- /:/host:ro,rslave
restart: unless-stopped
# Container metrics. Published on <hostIP>:8080. If 8080 is taken on this host,
# remap the left side (e.g. "8098:8080") and update the VMStaticScrape target.
cadvisor:
image: gcr.io/cadvisor/cadvisor:v0.49.1
command:
- --housekeeping_interval=30s
- --docker_only=true
ports:
- "8080:8080"
volumes:
- /:/rootfs:ro
- /var/run:/var/run:ro
- /sys:/sys:ro
- /var/lib/docker/:/var/lib/docker:ro
- /dev/disk/:/dev/disk:ro
devices:
- /dev/kmsg
privileged: true
restart: unless-stopped
# Ships every container's logs to VictoriaLogs (pushed over the tailnet).
vector:
image: timberio/vector:0.44.0-debian
volumes:
- /var/run/docker.sock:/var/run/docker.sock:ro
- ./vector.yaml:/etc/vector/vector.yaml:ro
environment:
VLOGS_ENDPOINT: ${VLOGS_ENDPOINT}
HOST_LABEL: ${HOST_LABEL}
restart: unless-stopped
@@ -0,0 +1,30 @@
# Vector: tail all Docker container logs on this host and ship them to VictoriaLogs
# via the Elasticsearch bulk API (the VM-recommended Vector sink). Endpoint + host
# label come from the environment (see .env / compose.yaml).
sources:
docker:
type: docker_logs
transforms:
enrich:
type: remap
inputs: [docker]
source: |
.host = get_env_var!("HOST_LABEL")
sinks:
vlogs:
type: elasticsearch
inputs: [enrich]
endpoints:
- "${VLOGS_ENDPOINT}" # e.g. https://vlogs.tail7e769.ts.net/insert/elasticsearch/
mode: bulk
api_version: v8
compression: gzip
healthcheck:
enabled: false
query:
_msg_field: message
_time_field: timestamp
# Group log streams by host + container so LogsQL stream filters work well.
_stream_fields: host,container_name
+23
View File
@@ -0,0 +1,23 @@
#!/usr/bin/env bash
# Load every *.json in this folder into Grafana as a sidecar dashboard ConfigMap.
# The Grafana chart's dashboard sidecar (sidecar.dashboards.enabled) watches for
# ConfigMaps labeled grafana_dashboard=1 in the monitoring namespace and imports them.
# Idempotent — re-run after adding/updating a dashboard JSON.
#
# The VictoriaMetrics board uses a `$ds` datasource variable that resolves to the
# default datasource (VictoriaMetrics), so no per-panel rewiring is needed.
set -euo pipefail
cd "$(dirname "$0")"
NS=monitoring
for f in *.json; do
[ -e "$f" ] || continue
name="grafana-dashboard-$(basename "$f" .json)"
echo "applying $name from $f"
kubectl create configmap "$name" \
--namespace "$NS" \
--from-file="$f" \
--dry-run=client -o yaml \
| kubectl label --local -f - grafana_dashboard=1 --dry-run=client -o yaml \
| kubectl apply -f -
done
File diff suppressed because it is too large Load Diff
+22
View File
@@ -0,0 +1,22 @@
#!/usr/bin/env bash
# Deploy Grafana. Create the OIDC client secret first (see oidc-secret.yaml header),
# then install the chart. Requires the metrics/logs/traces backends to exist so the
# provisioned datasources resolve.
set -euo pipefail
# 1) OIDC client secret (edit oidc-secret.yaml, or create it imperatively — preferred).
kubectl apply -f oidc-secret.yaml
helm repo add grafana https://grafana.github.io/helm-charts
helm repo update grafana
# Pin --version after the first install (helm search repo grafana/grafana --versions).
helm upgrade --install grafana grafana/grafana \
--namespace monitoring \
--values values.yaml \
--wait
# The tailnet FQDN, once the tailscale operator assigns it:
# kubectl -n monitoring get ingress grafana -o jsonpath='{.status.loadBalancer.ingress[0].hostname}'
# Break-glass admin password:
# kubectl -n monitoring get secret grafana -o jsonpath='{.data.admin-password}' | base64 -d
@@ -0,0 +1,18 @@
# Plaintext OIDC client secret Grafana presents to Authelia. Authelia stores only the
# pbkdf2-sha512 HASH of this same value (see authelia/values.yaml grafana client).
#
# Generate a matching pair:
# authelia crypto hash generate pbkdf2 --variant sha512 --random --random.length 72
# Put the "Random Password" (plaintext) below; put the "Digest" (hash) in Authelia.
#
# Do NOT commit the real secret. Create it out-of-band instead, e.g.:
# kubectl -n monitoring create secret generic grafana-oidc \
# --from-literal=client_secret='<plaintext>'
apiVersion: v1
kind: Secret
metadata:
name: grafana-oidc
namespace: monitoring
type: Opaque
stringData:
client_secret: "REPLACE_ME_WITH_PLAINTEXT_OIDC_SECRET"
@@ -0,0 +1,98 @@
# Grafana — single pane over metrics (VictoriaMetrics), logs (VictoriaLogs) and
# traces (VictoriaTraces via Jaeger API). Exposed privately on the tailnet
# (grafana.tail7e769.ts.net) and authenticated via Authelia OIDC. Local admin is
# break-glass only.
#
# Chart: grafana/grafana (repo: https://grafana.github.io/helm-charts)
# --- VictoriaLogs needs its Grafana datasource plugin ---
plugins:
- victoriametrics-logs-datasource
# --- Dashboard sidecar: auto-loads any ConfigMap labeled grafana_dashboard=1 in the
# namespace. Migrated boards live in ./dashboards and are applied by ./dashboards/apply.sh ---
sidecar:
dashboards:
enabled: true
label: grafana_dashboard
labelValue: "1"
folderAnnotation: grafana_folder
provider:
foldersFromFilesStructure: true
# --- Provisioned datasources ---
datasources:
datasources.yaml:
apiVersion: 1
datasources:
- name: VictoriaMetrics
type: prometheus
access: proxy
url: http://vmsingle-main.monitoring.svc:8428
isDefault: true
jsonData:
prometheusType: Prometheus
- name: VictoriaLogs
type: victoriametrics-logs-datasource
access: proxy
url: http://vlsingle-main.monitoring.svc:9428
- name: VictoriaTraces
type: jaeger
access: proxy
# VictoriaTraces (VTSingle CR) exposes a Jaeger-compatible query API under /select/jaeger.
url: http://vtsingle-main.monitoring.svc:10428/select/jaeger
# --- Persistence on OpenEBS ZFS ---
persistence:
enabled: true
storageClassName: localpv-zfs-ceph
size: 5Gi
# --- Private exposure via the Tailscale ingress (like seaweedfs-admin) ---
# The tailscale operator provisions grafana.<tailnet>.ts.net and a TLS cert.
ingress:
enabled: true
ingressClassName: tailscale
hosts:
- grafana
tls:
- hosts:
- grafana
# --- OIDC via Authelia (AD groups -> Grafana roles) ---
# client_secret is injected from the grafana-oidc Secret (see oidc-secret.yaml),
# which overrides any value in grafana.ini.
envValueFrom:
GF_AUTH_GENERIC_OAUTH_CLIENT_SECRET:
secretKeyRef:
name: grafana-oidc
key: client_secret
grafana.ini:
server:
root_url: "https://grafana.tail7e769.ts.net" # must match the tailnet FQDN + Authelia redirect_uri
auth:
# Keep the local admin login available as break-glass; don't force OIDC-only.
disable_login_form: false
oauth_auto_login: false
auth.generic_oauth:
enabled: true
name: Authelia
client_id: grafana
scopes: "openid profile email groups"
auth_url: "https://auth.ddupan.top/api/oidc/authorization"
token_url: "https://auth.ddupan.top/api/oidc/token"
api_url: "https://auth.ddupan.top/api/oidc/userinfo"
login_attribute_path: preferred_username
name_attribute_path: name
email_attribute_path: email
groups_attribute_path: groups
# AD "Enterprise Admins" -> full Grafana server admin; "Domain Admins" -> org Admin;
# everyone else who can authenticate -> Viewer. Tune group names to taste.
role_attribute_path: "contains(groups[*], 'Enterprise Admins') && 'GrafanaAdmin' || contains(groups[*], 'Domain Admins') && 'Admin' || 'Viewer'"
allow_assign_grafana_admin: true
role_attribute_strict: false
use_pkce: true
# Dashboards are loaded by the sidecar (above) from ConfigMaps created by
# ./dashboards/apply.sh. Drop more JSON into ./dashboards and re-run that script.
@@ -0,0 +1,22 @@
# victoria-logs-collector — deploys vlagent as a DaemonSet that discovers and tails
# every pod's container logs on each node and ships them to VictoriaLogs. Buffers to
# a local volume when VictoriaLogs is unavailable.
#
# NOTE: confirm key names against the installed chart before applying:
# helm show values vm/victoria-logs-collector
# (chart value layout occasionally shifts between versions.)
remoteWrite:
- url: "http://vlsingle-main.monitoring.svc:9428"
# Tolerate control-plane taints so logs are collected from every node.
tolerations:
- operator: Exists
resources:
requests:
cpu: 50m
memory: 128Mi
limits:
cpu: 500m
memory: 512Mi
+15
View File
@@ -0,0 +1,15 @@
#!/usr/bin/env bash
# Deploy the pod-log collector (vlagent DaemonSet) that ships container logs to the
# VLSingle instance created by ./vlsingle.yaml. Apply vlsingle.yaml first.
set -euo pipefail
kubectl apply -f vlsingle.yaml
helm repo add vm https://victoriametrics.github.io/helm-charts/
helm repo update vm
# Pin --version after the first install (helm search repo vm/victoria-logs-collector --versions).
helm upgrade --install victoria-logs-collector vm/victoria-logs-collector \
--namespace monitoring \
--values collector-values.yaml \
--wait
@@ -0,0 +1,24 @@
# Private tailnet exposure of the VictoriaLogs INGEST endpoint so off-cluster log
# shippers (Vector on Docker hosts) can push to it. Provisions vlogs.<tailnet>.ts.net.
# Ingestion is unauthenticated — acceptable on the private tailnet; do not expose publicly.
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: victorialogs
namespace: monitoring
spec:
ingressClassName: tailscale
rules:
- host: vlogs
http:
paths:
- path: /
pathType: Prefix
backend:
service:
name: vlsingle-main
port:
number: 9428
tls:
- hosts:
- vlogs
+25
View File
@@ -0,0 +1,25 @@
# Single-node VictoriaLogs, managed by the same VM operator as the metrics stack.
# Service exposed by the operator: vlsingle-main.monitoring.svc:9428
# Query via Grafana (VictoriaLogs datasource) with LogsQL.
# NOTE: the VictoriaLogs/Traces CRDs are served under .../v1 (metrics CRDs use v1beta1).
apiVersion: operator.victoriametrics.com/v1
kind: VLSingle
metadata:
name: main
namespace: monitoring
spec:
retentionPeriod: "30d"
storage:
accessModes:
- ReadWriteOnce
storageClassName: localpv-zfs-ceph
resources:
requests:
storage: 30Gi
resources:
requests:
cpu: 100m
memory: 256Mi
limits:
cpu: "1"
memory: 2Gi
@@ -0,0 +1,41 @@
# vmagent scrape-config hot-reload on k3s
## Symptom
New/changed `VMServiceScrape`/`VMNodeScrape`/`VMStaticScrape`/... are written by the
operator into the `vmagent-main` config Secret, but the running vmagent keeps serving
its **startup** config — the new targets never appear until vmagent is restarted.
## Cause
vmagent's config-reloader (`victoriametrics/operator:config-reloader-*`) watches the
config Secret with a client-go informer, which does `list` **and** `watch`. The
operator-generated Role `monitoring:monitoring:vmagent-main` grants `secrets: [get, watch]`
but **not `list`**, so on k3s the informer errors with:
```
cannot list resource "secrets" in API group "" in the namespace "monitoring"
```
and no reload is ever triggered.
## Fix options
**A) Grant `list` on secrets (proper hot-reload).** Additive, namespace-scoped Role
(vmagent already has get/watch; this adds only `list`). Apply `reload-rbac.yaml`:
```bash
kubectl apply -f metrics/reload-rbac.yaml
```
Marginal exposure: lets the vmagent SA enumerate Secrets in the `monitoring` namespace
only (it can already read them individually). This is what the operator is expected to
grant; the generated Role simply omits `list` on this version.
**B) No RBAC change — restart on change.** Leave RBAC as-is and restart vmagent after
applying scrape-config changes:
```bash
kubectl -n monitoring rollout restart deploy/vmagent-main
```
Fine for a homelab where scrape configs change rarely; the tradeoff is a manual step
(and a ~10s gap) each time.
@@ -0,0 +1,20 @@
#!/usr/bin/env bash
# kube-state-metrics — metrics about k8s objects (deployments, pods, nodes, PVCs, ...).
# Its chart ships a ServiceMonitor; the VM operator's Prometheus converter turns that
# into a VMServiceScrape automatically, so no hand-written scrape CR is needed here.
#
# Prereq: the Prometheus-operator CRDs (ServiceMonitor, ...) must exist in the cluster
# for `serviceMonitor.enabled` to apply. If they don't:
# kubectl apply --server-side -f \
# https://github.com/prometheus-operator/prometheus-operator/releases/latest/download/bundle.yaml \
# # (or just the monitoring.coreos.com CRDs)
# Alternatively, disable the ServiceMonitor below and add a VMServiceScrape by hand.
set -euo pipefail
helm repo add prometheus-community https://prometheus-community.github.io/helm-charts
helm repo update prometheus-community
helm upgrade --install kube-state-metrics prometheus-community/kube-state-metrics \
--namespace monitoring \
--set prometheus.monitor.enabled=true \
--wait
@@ -0,0 +1,67 @@
# Host metrics (CPU/mem/disk/net/filesystem) for every node. node-exporter has no
# ServiceMonitor of its own here, so we scrape it directly with a VMNodeScrape:
# vmagent discovers each Node and scrapes its InternalIP:9100.
apiVersion: apps/v1
kind: DaemonSet
metadata:
name: node-exporter
namespace: monitoring
labels:
app.kubernetes.io/name: node-exporter
spec:
selector:
matchLabels:
app.kubernetes.io/name: node-exporter
template:
metadata:
labels:
app.kubernetes.io/name: node-exporter
spec:
hostNetwork: true # bind 9100 on the node so VMNodeScrape can reach it
hostPID: true
tolerations:
- operator: Exists # run on control-plane / tainted nodes too
containers:
- name: node-exporter
image: quay.io/prometheus/node-exporter:v1.9.1
args:
- --path.procfs=/host/proc
- --path.sysfs=/host/sys
- --path.rootfs=/host/root
- --collector.filesystem.mount-points-exclude=^/(dev|proc|sys|var/lib/docker/.+|var/lib/kubelet/.+)($|/)
ports:
- name: metrics
containerPort: 9100
hostPort: 9100
resources:
requests:
cpu: 25m
memory: 32Mi
limits:
cpu: 200m
memory: 128Mi
volumeMounts:
- { name: proc, mountPath: /host/proc, readOnly: true }
- { name: sys, mountPath: /host/sys, readOnly: true }
- { name: root, mountPath: /host/root, readOnly: true, mountPropagation: HostToContainer }
volumes:
- { name: proc, hostPath: { path: /proc } }
- { name: sys, hostPath: { path: /sys } }
- { name: root, hostPath: { path: / } }
---
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMNodeScrape
metadata:
name: node-exporter
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
port: "9100"
scheme: http
interval: 30s
relabelConfigs:
- action: labelmap
regex: __meta_kubernetes_node_label_(.+)
- targetLabel: job
replacement: node-exporter
@@ -0,0 +1,29 @@
# Enables vmagent scrape-config hot-reload. The operator-generated Role grants the
# vmagent-main SA secrets [get,watch] but omits [list], which its config-reloader's
# informer requires — so scrape-config changes stall until vmagent restarts.
# This supplementary, namespace-scoped Role covers secrets get/list/watch so the
# reloader works regardless of the operator Role. Scope: Secrets in `monitoring` only.
# See README-reload.md for background.
apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
name: vmagent-main-secret-lister
namespace: monitoring
rules:
- apiGroups: [""]
resources: ["secrets"]
verbs: ["get", "list", "watch"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
name: vmagent-main-secret-lister
namespace: monitoring
roleRef:
apiGroup: rbac.authorization.k8s.io
kind: Role
name: vmagent-main-secret-lister
subjects:
- kind: ServiceAccount
name: vmagent-main
namespace: monitoring
@@ -0,0 +1,127 @@
# Ported from ../../../../apps/victoriametrics/rules/alerts-health.yml
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMRule
metadata:
name: vm-health
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
groups:
- name: vm-health
rules:
- alert: TooManyRestarts
expr: changes(process_start_time_seconds{job=~".*(victoriametrics|vmselect|vminsert|vmstorage|vmagent|vmalert|vmsingle|vmalertmanager|vmauth).*"}[15m]) > 2
labels:
severity: critical
annotations:
summary: "{{ $labels.job }} too many restarts (instance {{ $labels.instance }})"
description: >
Job {{ $labels.job }} (instance {{ $labels.instance }}) has restarted more than twice in the last 15 minutes.
It might be crashlooping.
- alert: ServiceDown
expr: up{job=~".*(victoriametrics|vmselect|vminsert|vmstorage|vmagent|vmalert|vmsingle|vmalertmanager|vmauth).*"} == 0
for: 2m
labels:
severity: critical
annotations:
summary: "Service {{ $labels.job }} is down on {{ $labels.instance }}"
description: "{{ $labels.instance }} of job {{ $labels.job }} has been down for more than 2 minutes."
- alert: ProcessNearFDLimits
expr: (process_max_fds - process_open_fds) < 100
for: 5m
labels:
severity: critical
annotations:
summary: "Number of free file descriptors is less than 100 for \"{{ $labels.job }}\"(\"{{ $labels.instance }}\") for the last 5m"
description: |
Exhausting OS file descriptors limit can cause severe degradation of the process.
Consider to increase the limit as fast as possible.
- alert: TooHighMemoryUsage
expr: (min_over_time(process_resident_memory_anon_bytes[10m]) / vm_available_memory_bytes) > 0.8
for: 5m
labels:
severity: critical
annotations:
summary: "It is more than 80% of memory used by \"{{ $labels.job }}\"(\"{{ $labels.instance }}\")"
description: |
Too high memory usage may result into multiple issues such as OOMs or degraded performance.
Consider to either increase available memory or decrease the load on the process.
- alert: TooHighCPUUsage
expr: rate(process_cpu_seconds_total[5m]) / process_cpu_cores_available > 0.9
for: 5m
labels:
severity: critical
annotations:
summary: "More than 90% of CPU is used by \"{{ $labels.job }}\"(\"{{ $labels.instance }}\") during the last 5m"
description: >
Too high CPU usage may be a sign of insufficient resources and make process unstable.
Consider to either increase available CPU resources or decrease the load on the process.
- alert: TooHighGoroutineSchedulingLatency
expr: histogram_quantile(0.99, sum(rate(go_sched_latencies_seconds_bucket{job=~".*(victoriametrics|vmselect|vminsert|vmstorage|vmagent|vmalert|vmsingle|vmalertmanager|vmauth).*"}[5m])) by (le, job, instance)) > 0.1
for: 15m
labels:
severity: critical
annotations:
summary: "\"{{ $labels.job }}\"(\"{{ $labels.instance }}\") has insufficient CPU resources for >15m"
description: >
Go runtime is unable to schedule goroutines execution in acceptable time. This is usually a sign of
insufficient CPU resources or CPU throttling.
- alert: TooManyLogs
expr: sum(increase(vm_log_messages_total{level="error"}[5m])) without (app_version, location) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Too many logs printed for job \"{{ $labels.job }}\" ({{ $labels.instance }})"
description: >
Logging rate for job \"{{ $labels.job }}\" ({{ $labels.instance }}) is {{ $value }} for last 15m.
Worth to check logs for specific error messages.
- alert: TooManyTSIDMisses
expr: increase(vm_missing_tsids_for_metric_id_total[5m]) > 0
for: 15m
labels:
severity: critical
annotations:
summary: "Unexpected TSID misses for job \"{{ $labels.job }}\" ({{ $labels.instance }}) for the last 15 minutes"
description: |
Unexpected TSID misses for \"{{ $labels.job }}\" ({{ $labels.instance }}) for the last 15 minutes.
If this happens after unclean shutdown of VictoriaMetrics process (via \"kill -9\", OOM or power off),
then this is OK - the alert must go away in a few minutes after the restart.
- alert: ConcurrentInsertsHitTheLimit
expr: avg_over_time(vm_concurrent_insert_current[1m]) >= vm_concurrent_insert_capacity
for: 15m
labels:
severity: warning
annotations:
summary: "{{ $labels.job }} on instance {{ $labels.instance }} is constantly hitting concurrent inserts limit"
description: |
The limit of concurrent inserts on instance {{ $labels.instance }} depends on the number of CPUs.
Usually, when component constantly hits the limit it is likely the component is overloaded and requires more CPU.
- alert: IndexDBRecordsDrop
expr: increase(vm_indexdb_items_dropped_total[5m]) > 0
labels:
severity: critical
annotations:
summary: "IndexDB skipped registering items during data ingestion with reason={{ $labels.reason }}."
description: |
VictoriaMetrics could skip registering new timeseries during ingestion if they fail the validation process.
- alert: RowsRejectedOnIngestion
expr: rate(vm_rows_ignored_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Some rows are rejected on \"{{ $labels.instance }}\" on ingestion attempt"
description: "Ingested rows on instance \"{{ $labels.instance }}\" are rejected due to the
following reason: \"{{ $labels.reason }}\""
- alert: TooHighQueryLoad
expr: increase(vm_concurrent_select_limit_timeout_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Read queries fail with timeout for {{ $labels.job }} on instance {{ $labels.instance }}"
description: |
Instance {{ $labels.instance }} ({{ $labels.job }}) is failing to serve read queries during last 15m.
Possible solutions: reduce the query load; increase compute resources; adjust search concurrency limits.
@@ -0,0 +1,137 @@
# Ported from ../../../../apps/victoriametrics/rules/alerts-vmagent.yml
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMRule
metadata:
name: vmagent
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
groups:
- name: vmagent
interval: 30s
concurrency: 2
rules:
- alert: PersistentQueueIsDroppingData
expr: sum(increase(vm_persistentqueue_bytes_dropped_total[5m])) without (path) > 0
for: 10m
labels:
severity: critical
annotations:
summary: "Instance {{ $labels.instance }} is dropping data from persistent queue"
description: "Vmagent dropped {{ $value | humanize1024 }} from persistent queue
on instance {{ $labels.instance }} for the last 10m."
- alert: RejectedRemoteWriteDataBlocksAreDropped
expr: sum(increase(vmagent_remotewrite_packets_dropped_total[5m])) without (url) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Vmagent is dropping data blocks that are rejected by remote storage"
description: "Job \"{{ $labels.job }}\" on instance {{ $labels.instance }} drops the rejected by
remote-write server data blocks. Check the logs to find the reason for rejects."
- alert: TooManyScrapeErrors
expr: increase(vm_promscrape_scrapes_failed_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Vmagent fails to scrape one or more targets"
description: "Job \"{{ $labels.job }}\" on instance {{ $labels.instance }} fails to scrape targets for last 15m"
- alert: ScrapePoolHasNoTargets
expr: sum(vm_promscrape_scrape_pool_targets) without (status, instance, pod) == 0
for: 30m
labels:
severity: warning
annotations:
summary: "Vmagent has scrape_pool with 0 configured/discovered targets"
description: "Vmagent \"{{ $labels.job }}\" has scrape_pool \"{{ $labels.scrape_job }}\"
with 0 discovered targets. It is likely a misconfiguration."
- alert: TooManyWriteErrors
expr: |
(sum(increase(vm_ingestserver_request_errors_total[5m])) without (name,net,type)
+
sum(increase(vmagent_http_request_errors_total[5m])) without (path,protocol)) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Vmagent responds with too many errors on data ingestion protocols"
description: "Job \"{{ $labels.job }}\" on instance {{ $labels.instance }} responds with errors to write requests for last 15m."
- alert: TooManyRemoteWriteErrors
expr: rate(vmagent_remotewrite_retries_count_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Job \"{{ $labels.job }}\" on instance {{ $labels.instance }} fails to push to remote storage"
description: "Vmagent fails to push data via remote write protocol to destination \"{{ $labels.url }}\".
Ensure that destination is up and reachable."
- alert: RemoteWriteConnectionIsSaturated
expr: |
(
rate(vmagent_remotewrite_send_duration_seconds_total[5m])
/
vmagent_remotewrite_queues
) > 0.9
for: 15m
labels:
severity: warning
annotations:
summary: "Remote write connection from \"{{ $labels.job }}\" (instance {{ $labels.instance }}) to {{ $labels.url }} is saturated"
description: "The remote write connection between vmagent and destination \"{{ $labels.url }}\"
is saturated by more than 90%. Increase -remoteWrite.queues or check the destination's capacity."
- alert: PersistentQueueForWritesIsSaturated
expr: rate(vm_persistentqueue_write_duration_seconds_total[5m]) > 0.9
for: 15m
labels:
severity: warning
annotations:
summary: "Persistent queue writes for instance {{ $labels.instance }} are saturated"
description: "Persistent queue writes are saturated by more than 90%. Reduce load or improve disk throughput."
- alert: PersistentQueueForReadsIsSaturated
expr: rate(vm_persistentqueue_read_duration_seconds_total[5m]) > 0.9
for: 15m
labels:
severity: warning
annotations:
summary: "Persistent queue reads for instance {{ $labels.instance }} are saturated"
description: "Persistent queue reads are saturated by more than 90%. Reduce load or improve disk throughput."
- alert: SeriesLimitHourReached
expr: (vmagent_hourly_series_limit_current_series / vmagent_hourly_series_limit_max_series) > 0.9
labels:
severity: critical
annotations:
summary: "Instance {{ $labels.instance }} reached 90% of the hourly series limit"
description: "Max series limit set via -remoteWrite.maxHourlySeries is close to the max value."
- alert: SeriesLimitDayReached
expr: (vmagent_daily_series_limit_current_series / vmagent_daily_series_limit_max_series) > 0.9
labels:
severity: critical
annotations:
summary: "Instance {{ $labels.instance }} reached 90% of the daily series limit"
description: "Max series limit set via -remoteWrite.maxDailySeries is close to the max value."
- alert: ConfigurationReloadFailure
expr: |
vm_promscrape_config_last_reload_successful != 1
or
vmagent_relabel_config_last_reload_successful != 1
labels:
severity: warning
annotations:
summary: "Configuration reload failed for vmagent instance {{ $labels.instance }}"
description: "Configuration hot-reload failed for vmagent on instance {{ $labels.instance }}. Check the logs."
- alert: StreamAggrFlushTimeout
expr: increase(vm_streamaggr_flush_timeouts_total[5m]) > 0
labels:
severity: warning
annotations:
summary: "Streaming aggregation at \"{{ $labels.job }}\" (instance {{ $labels.instance }}) can't keep up."
description: "Stream aggregation can't finish within the configured interval and may produce incorrect results."
- alert: StreamAggrDedupFlushTimeout
expr: increase(vm_streamaggr_dedup_flush_timeouts_total[5m]) > 0
labels:
severity: warning
annotations:
summary: "Deduplication \"{{ $labels.job }}\" (instance {{ $labels.instance }}) can't keep up."
description: "Deduplication can't finish within the configured interval and may produce incorrect results."
@@ -0,0 +1,77 @@
# Ported from ../../../../apps/victoriametrics/rules/alerts-vmalert.yml
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMRule
metadata:
name: vmalert
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
groups:
- name: vmalert
interval: 30s
rules:
- alert: ConfigurationReloadFailure
expr: vmalert_config_last_reload_successful != 1
labels:
severity: warning
annotations:
summary: "Configuration reload failed for vmalert instance {{ $labels.instance }}"
description: "Configuration hot-reload failed for vmalert on instance {{ $labels.instance }}. Check the logs."
- alert: AlertingRulesError
expr: sum(increase(vmalert_alerting_rules_errors_total[5m])) without(id) > 0
for: 5m
labels:
severity: warning
annotations:
summary: "Alerting rules are failing for vmalert instance {{ $labels.instance }}"
description: "Alerting rules execution is failing for group \"{{ $labels.group }}\" in file \"{{ $labels.file }}\". Check the logs."
- alert: RecordingRulesError
expr: sum(increase(vmalert_recording_rules_errors_total[5m])) without(id) > 0
for: 5m
labels:
severity: warning
annotations:
summary: "Recording rules are failing for vmalert instance {{ $labels.instance }}"
description: "Recording rules execution is failing for group \"{{ $labels.group }}\" in file \"{{ $labels.file }}\". Check the logs."
- alert: RecordingRulesNoData
expr: sum(vmalert_recording_rules_last_evaluation_samples) without(id) < 1
for: 30m
labels:
severity: info
annotations:
summary: "Recording rule ({{ $labels.group }}) produces no data"
description: "Recording rule from group \"{{ $labels.group }}\" in file \"{{ $labels.file }}\"
produces 0 samples over the last 30min. Possible misconfiguration or incorrect query."
- alert: TooManyMissedIterations
expr: increase(vmalert_iteration_missed_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "vmalert instance {{ $labels.instance }} is missing rules evaluations"
description: "Group evaluation takes longer than the configured interval. Increase interval or concurrency."
- alert: RemoteWriteErrors
expr: increase(vmalert_remotewrite_errors_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "vmalert instance {{ $labels.instance }} is failing to push metrics to remote write URL"
description: "vmalert is failing to push alerting/recording rule metrics to remote write. Check the logs."
- alert: RemoteWriteDroppingData
expr: increase(vmalert_remotewrite_dropped_rows_total[5m]) > 0
for: 5m
labels:
severity: critical
annotations:
summary: "vmalert instance {{ $labels.instance }} is dropping data sent to remote write URL"
description: "vmalert is dropping alerting/recording rule results. This may cause gaps in state. Check the logs."
- alert: AlertmanagerErrors
expr: increase(vmalert_alerts_send_errors_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "vmalert instance {{ $labels.instance }} is failing to send notifications to Alertmanager"
description: "vmalert is failing to send alert notifications to \"{{ $labels.addr }}\". Check the logs."
@@ -0,0 +1,127 @@
# Ported from ../../../../apps/victoriametrics/rules/alerts.yml
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMRule
metadata:
name: vmsingle
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
groups:
- name: vmsingle
interval: 30s
concurrency: 2
rules:
- alert: DiskRunsOutOfSpaceIn3Days
expr: |
sum(vm_free_disk_space_bytes) without(path) /
(
(rate(vm_rows_added_to_storage_total[1d]) - sum(rate(vm_deduplicated_samples_total[1d])) without(type)) * (
sum(vm_data_size_bytes{type!~"indexdb.*"}) without(type) /
sum(vm_rows{type!~"indexdb.*"}) without(type)
)
+
rate(vm_new_timeseries_created_total[1d]) * (
sum(vm_data_size_bytes{type="indexdb/file"}) without(type)/
sum(vm_rows{type="indexdb/file"}) without(type)
)
) < 3 * 24 * 3600 > 0
for: 30m
labels:
severity: critical
annotations:
summary: "Instance {{ $labels.instance }} will run out of disk space soon"
description: "Taking into account current ingestion rate, free disk space will be enough only
for {{ $value | humanizeDuration }} on instance {{ $labels.instance }}.\n
Consider to limit the ingestion rate, decrease retention or scale the disk space if possible."
- alert: NodeBecomesReadonlyIn3Days
expr: |
sum(vm_free_disk_space_bytes - vm_free_disk_space_limit_bytes) without(path) /
(
(rate(vm_rows_added_to_storage_total[1d]) - sum(rate(vm_deduplicated_samples_total[1d])) without(type)) * (
sum(vm_data_size_bytes{type!~"indexdb.*"}) without(type) /
sum(vm_rows{type!~"indexdb.*"}) without(type)
)
+
rate(vm_new_timeseries_created_total[1d]) * (
sum(vm_data_size_bytes{type="indexdb/file"}) without(type) /
sum(vm_rows{type="indexdb/file"}) without(type)
)
) < 3 * 24 * 3600 > 0
for: 30m
labels:
severity: warning
annotations:
summary: "Instance {{ $labels.instance }} will become read-only in 3 days"
description: "Taking into account current ingestion rate and free disk space
instance {{ $labels.instance }} is writable for {{ $value | humanizeDuration }}.\n
Consider to limit the ingestion rate, decrease retention or scale the disk space up if possible."
- alert: DiskRunsOutOfSpace
expr: |
sum(vm_data_size_bytes) by(job, instance) /
(
sum(vm_free_disk_space_bytes) by(job, instance) +
sum(vm_data_size_bytes) by(job, instance)
) > 0.8
for: 30m
labels:
severity: critical
annotations:
summary: "Instance {{ $labels.instance }} (job={{ $labels.job }}) will run out of disk space soon"
description: "Disk utilisation on instance {{ $labels.instance }} is more than 80%.\n
Having less than 20% of free disk space could cripple merge processes and overall performance.
Consider to limit the ingestion rate, decrease retention or scale the disk space if possible."
- alert: RequestErrorsToAPI
expr: increase(vm_http_request_errors_total[5m]) > 0
for: 15m
labels:
severity: warning
annotations:
summary: "Too many errors served for path {{ $labels.path }} (instance {{ $labels.instance }})"
description: "Requests to path {{ $labels.path }} are receiving errors.
Please verify if clients are sending correct requests."
- alert: TooHighChurnRate
expr: |
(
sum(rate(vm_new_timeseries_created_total[5m])) by(instance)
/
sum(rate(vm_rows_inserted_total[5m])) by(instance)
) > 0.1
for: 15m
labels:
severity: warning
annotations:
summary: "Churn rate is more than 10% on \"{{ $labels.instance }}\" for the last 15m"
description: "VM constantly creates new time series on \"{{ $labels.instance }}\".\n
This effect is known as Churn Rate.\n
High Churn Rate is tightly connected with database performance and may
result in unexpected OOM's or slow queries."
- alert: TooHighChurnRate24h
expr: |
sum(increase(vm_new_timeseries_created_total[24h])) by(instance)
>
(sum(vm_cache_entries{type="storage/hour_metric_ids"}) by(instance) * 3)
for: 15m
labels:
severity: warning
annotations:
summary: "Too high number of new series on \"{{ $labels.instance }}\" created over last 24h"
description: "The number of created new time series over last 24h is 3x times higher than
current number of active series on \"{{ $labels.instance }}\".\n
This effect is known as Churn Rate.\n
High Churn Rate is tightly connected with database performance and may
result in unexpected OOM's or slow queries."
- alert: TooHighSlowInsertsRate
expr: |
(
sum(rate(vm_slow_row_inserts_total[5m])) by(instance)
/
sum(rate(vm_rows_inserted_total[5m])) by(instance)
) > 0.05
for: 15m
labels:
severity: warning
annotations:
summary: "Percentage of slow inserts is more than 5% on \"{{ $labels.instance }}\" for the last 15m"
description: "High rate of slow inserts on \"{{ $labels.instance }}\" may be a sign of resource exhaustion
for the current load. It is likely more RAM is needed for optimal handling of the current number of active time series."
@@ -0,0 +1,30 @@
# Blocky DNS metrics. Same VMStaticScrape pattern as docker-hosts.yaml, because
# Blocky runs as a compose stack on the laptop rather than in the cluster
# (see ../../../../apps/blocky/README.md for why DNS deliberately lives outside k3s).
#
# ⚠ Blocky is STAGED, not deployed — this target will be DOWN until it is started.
# That is expected; do not go hunting for a broken exporter.
#
# Worth alerting on once it is live, because there is currently no DNS visibility
# at all:
# blocky_error_total — upstream failures
# blocky_denylist_cache_entries — collapses to 0 if a list fetch fails
# blocky_cache_hits_total / _misses_total — cache effectiveness
# blocky_query_total{response_type=...} — BLOCKED vs CACHED vs RESOLVED split
---
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMStaticScrape
metadata:
name: blocky
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
jobName: blocky
targetEndpoints:
- targets:
- "192.168.10.127:4000"
path: /metrics
labels:
job: blocky
host: docker-01
@@ -0,0 +1,29 @@
# Pull metrics from Docker hosts running ../../docker-hosts/compose.yaml.
# vmagent scrapes each host's node-exporter (:9100) and cAdvisor (:8080) over the LAN.
# Add one target block per host; keep the `host` label in sync with its HOST_LABEL.
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMStaticScrape
metadata:
name: docker-hosts
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
jobName: docker-hosts
targetEndpoints:
# --- docker-01 (192.168.10.127) ---
- targets:
- "192.168.10.127:9100"
labels:
job: node-exporter
host: docker-01
- targets:
- "192.168.10.127:8080"
labels:
job: cadvisor
host: docker-01
# --- add more hosts below, mirroring the two blocks above ---
# - targets: ["192.168.10.x:9100"]
# labels: { job: node-exporter, host: docker-02 }
# - targets: ["192.168.10.x:8080"]
# labels: { job: cadvisor, host: docker-02 }
@@ -0,0 +1,47 @@
# Kubelet + cAdvisor metrics (per-node pod/container CPU, memory, network, fs).
# Scraped over the kubelet's authenticated https port using the vmagent pod's
# ServiceAccount token. Pattern from the operator docs (VMNodeScrape / cadvisor).
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMNodeScrape
metadata:
name: kubelet
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
scheme: https
honorLabels: true
honorTimestamps: false
interval: 30s
tlsConfig:
insecureSkipVerify: true
caFile: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt
bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
relabelConfigs:
- action: labelmap
regex: __meta_kubernetes_node_label_(.+)
- targetLabel: job
replacement: kubelet
---
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMNodeScrape
metadata:
name: cadvisor
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
scheme: https
honorLabels: true
honorTimestamps: false
interval: 30s
path: /metrics/cadvisor
tlsConfig:
insecureSkipVerify: true
caFile: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt
bearerTokenFile: /var/run/secrets/kubernetes.io/serviceaccount/token
relabelConfigs:
- action: labelmap
regex: __meta_kubernetes_node_label_(.+)
- targetLabel: job
replacement: cadvisor
@@ -0,0 +1,21 @@
# Scrape agent. selectAllByDefault picks up every VMServiceScrape / VMPodScrape /
# VMNodeScrape / VMStaticScrape in ALL namespaces, so app teams add their own
# scrape CRs and vmagent discovers them automatically. Writes to VMSingle.
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMAgent
metadata:
name: main
namespace: monitoring
spec:
replicaCount: 1
selectAllByDefault: true
scrapeInterval: 30s
remoteWrite:
- url: http://vmsingle-main.monitoring.svc:8428/api/v1/write
resources:
requests:
cpu: 100m
memory: 256Mi
limits:
cpu: "1"
memory: 1Gi
@@ -0,0 +1,27 @@
# Rule evaluator. selectAllByDefault picks up every VMRule in all namespaces
# (see ./rules). Reads from and writes alert/recording state to VMSingle;
# sends firing alerts to VMAlertmanager.
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMAlert
metadata:
name: main
namespace: monitoring
spec:
replicaCount: 1
selectAllByDefault: true
evaluationInterval: 30s
datasource:
url: http://vmsingle-main.monitoring.svc:8428
remoteWrite:
url: http://vmsingle-main.monitoring.svc:8428
remoteRead:
url: http://vmsingle-main.monitoring.svc:8428
notifier:
url: http://vmalertmanager-main.monitoring.svc:9093
resources:
requests:
cpu: 50m
memory: 128Mi
limits:
cpu: 500m
memory: 512Mi
@@ -0,0 +1,30 @@
# Alert router/notifier. Ported 1:1 from ../../../apps/victoriametrics/alertmanager.yaml,
# which currently blackholes everything. Wire real receivers here (email via the
# in-cluster smtp-relay, or a webhook) when you want notifications.
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMAlertmanager
metadata:
name: main
namespace: monitoring
spec:
replicaCount: 1
configRawYaml: |
route:
receiver: blackhole
receivers:
- name: blackhole
# Example email receiver via the in-cluster Postfix relay (smtp-relay/):
# receivers:
# - name: email
# email_configs:
# - to: '[email protected]'
# from: 'Alertmanager <[email protected]>'
# smarthost: 'smtp-relay.smtp-relay.svc.cluster.local:25'
# require_tls: false
resources:
requests:
cpu: 25m
memory: 64Mi
limits:
cpu: 200m
memory: 256Mi
@@ -0,0 +1,26 @@
# Single-node metrics TSDB. Replaces the Compose `victoriametrics` service.
# Service exposed by the operator: vmsingle-main.monitoring.svc:8428
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMSingle
metadata:
name: main
namespace: monitoring
spec:
retentionPeriod: "6" # months
removePvcAfterDelete: false # keep data if the CR is deleted
storage:
accessModes:
- ReadWriteOnce
storageClassName: localpv-zfs-ceph # OpenEBS ZFS (data/ceph pool)
resources:
requests:
storage: 50Gi
extraArgs:
dedup.minScrapeInterval: 30s
resources:
requests:
cpu: 250m
memory: 512Mi
limits:
cpu: "2"
memory: 4Gi
+6
View File
@@ -0,0 +1,6 @@
apiVersion: v1
kind: Namespace
metadata:
name: monitoring
labels:
app.kubernetes.io/part-of: observability
+20
View File
@@ -0,0 +1,20 @@
#!/usr/bin/env bash
# Install / upgrade the VictoriaMetrics operator into the monitoring namespace.
# The operator owns the VM* + VLSingle CRDs that everything else in this folder relies on.
set -euo pipefail
kubectl apply -f ../namespace.yaml
helm repo add vm https://victoriametrics.github.io/helm-charts/
helm repo update vm
# Pin the chart version after the first install for reproducibility:
# helm search repo vm/victoria-metrics-operator --versions | head
# then add: --version <x.y.z>
helm upgrade --install vm-operator vm/victoria-metrics-operator \
--namespace monitoring \
--values values.yaml \
--wait
# CRDs land cluster-wide; verify:
# kubectl get crd | grep victoriametrics
@@ -0,0 +1,38 @@
# victoria-metrics-operator — manages the VM* CRDs (VMSingle, VMAgent, VMAlert,
# VMAlertmanager, VMRule, VMServiceScrape, VMNodeScrape, ...) AND VLSingle (VictoriaLogs).
# One operator reconciles the whole metrics + logs stack declaratively.
#
# Install: ./helm.sh (chart: vm/victoria-metrics-operator)
# Let the chart install/upgrade the CRDs.
crds:
plain: true
cleanup:
enabled: false # keep CRDs (and thus CRs) if the operator is uninstalled
# Watch every namespace so app-owned VMServiceScrape/VMPodScrape/VMRule are picked up.
operator:
# convert legacy Prometheus-operator CRs too, harmless if none exist
disable_prometheus_converter: false
# Pin VictoriaMetrics' native config-reloader (this is already the operator default).
# NOTE: this reloader watches the config Secret via a client-go informer (list+watch).
# The operator-generated vmagent Role grants secrets [get,watch] but NOT [list], so on
# k3s the informer fails and scrape-config hot-reload stalls until vmagent is restarted.
# See ../metrics/README-reload.md for the fix options.
env:
- name: VM_USECUSTOMCONFIGRELOADER
value: "true"
# The operator only reconciles CRs; the actual components (vmsingle, vmagent, ...) are
# defined as CRs under ../metrics and ../logs. Keep the operator itself lean.
resources:
requests:
cpu: 50m
memory: 150Mi
limits:
cpu: 500m
memory: 500Mi
serviceMonitor:
enabled: true # let the operator self-scrape via a VMServiceScrape
+13
View File
@@ -0,0 +1,13 @@
#!/usr/bin/env bash
# Deploy tracing: VictoriaTraces (VTSingle CR, operator-managed) + the OpenTelemetry
# Collector that fronts it. Apps send OTLP to otel-collector.monitoring.svc:4317 (gRPC)
# / :4318 (HTTP); the collector batches and forwards to VictoriaTraces.
#
# NOTE: we use the operator's VTSingle CRD (traces/vtsingle.yaml) rather than the
# victoria-traces-single Helm chart — one operator manages metrics, logs, and traces.
# traces-values.yaml is kept only as a reference for the standalone-chart alternative.
set -euo pipefail
kubectl apply -f vtsingle.yaml
kubectl -n monitoring wait --for=jsonpath='{.status.updateStatus}'=operational vtsingle/main --timeout=120s || true
kubectl apply -f otel-collector.yaml
@@ -0,0 +1,117 @@
# OpenTelemetry Collector — OTLP ingress point for app traces. Receives OTLP over
# gRPC (4317) and HTTP (4318), batches, and exports to VictoriaTraces' OTLP endpoint.
# Point your apps' OTEL_EXPORTER_OTLP_ENDPOINT at otel-collector.monitoring.svc:4317.
apiVersion: v1
kind: ServiceAccount
metadata:
name: otel-collector
namespace: monitoring
---
apiVersion: v1
kind: ConfigMap
metadata:
name: otel-collector-config
namespace: monitoring
data:
config.yaml: |
receivers:
otlp:
protocols:
grpc:
endpoint: 0.0.0.0:4317
http:
endpoint: 0.0.0.0:4318
processors:
memory_limiter:
check_interval: 5s
limit_percentage: 80
spike_limit_percentage: 25
batch:
timeout: 5s
exporters:
otlphttp/victoriatraces:
# VictoriaTraces (VTSingle CR) OTLP-over-HTTP ingestion endpoint.
traces_endpoint: http://vtsingle-main.monitoring.svc:10428/insert/opentelemetry/v1/traces
tls:
insecure: true
service:
pipelines:
traces:
receivers: [otlp]
processors: [memory_limiter, batch]
exporters: [otlphttp/victoriatraces]
telemetry:
metrics:
address: 0.0.0.0:8888
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: otel-collector
namespace: monitoring
labels:
app.kubernetes.io/name: otel-collector
spec:
replicas: 1
selector:
matchLabels:
app.kubernetes.io/name: otel-collector
template:
metadata:
labels:
app.kubernetes.io/name: otel-collector
spec:
serviceAccountName: otel-collector
containers:
- name: otel-collector
# Pin to a current release; check https://github.com/open-telemetry/opentelemetry-collector-releases
image: otel/opentelemetry-collector-contrib:0.119.0
args: ["--config=/etc/otel/config.yaml"]
ports:
- { name: otlp-grpc, containerPort: 4317 }
- { name: otlp-http, containerPort: 4318 }
- { name: metrics, containerPort: 8888 }
volumeMounts:
- name: config
mountPath: /etc/otel
resources:
requests:
cpu: 50m
memory: 128Mi
limits:
cpu: 500m
memory: 512Mi
volumes:
- name: config
configMap:
name: otel-collector-config
---
apiVersion: v1
kind: Service
metadata:
name: otel-collector
namespace: monitoring
labels:
app.kubernetes.io/name: otel-collector
spec:
selector:
app.kubernetes.io/name: otel-collector
ports:
- { name: otlp-grpc, port: 4317, targetPort: 4317 }
- { name: otlp-http, port: 4318, targetPort: 4318 }
- { name: metrics, port: 8888, targetPort: 8888 }
---
# Self-scrape the collector's own telemetry into VictoriaMetrics.
apiVersion: operator.victoriametrics.com/v1beta1
kind: VMServiceScrape
metadata:
name: otel-collector
namespace: monitoring
labels:
app.kubernetes.io/part-of: victoria-metrics
spec:
selector:
matchLabels:
app.kubernetes.io/name: otel-collector
endpoints:
- port: metrics
@@ -0,0 +1,21 @@
# victoria-traces-single — single-node distributed tracing backend. No external
# storage deps. Ingests OTLP (from the OTel Collector) and exposes a Jaeger Query
# API that Grafana reads via a Jaeger datasource.
# Service: victoria-traces-single-server.monitoring.svc:10428
#
# NOTE: confirm key names against the installed chart before applying:
# helm show values vm/victoria-traces-single
server:
retentionPeriod: 30d
persistentVolume:
enabled: true
storageClassName: localpv-zfs-ceph
size: 20Gi
resources:
requests:
cpu: 100m
memory: 256Mi
limits:
cpu: "1"
memory: 2Gi
@@ -0,0 +1,25 @@
# Single-node VictoriaTraces, managed by the VM operator (no separate Helm chart needed
# — the operator ships the VTSingle CRD). Ingests OTLP from the OTel Collector and
# exposes a Jaeger-compatible query API that Grafana reads.
# Service: vtsingle-main.monitoring.svc:10428
apiVersion: operator.victoriametrics.com/v1
kind: VTSingle
metadata:
name: main
namespace: monitoring
spec:
retentionPeriod: "30d"
storage:
accessModes:
- ReadWriteOnce
storageClassName: localpv-zfs-ceph
resources:
requests:
storage: 20Gi
resources:
requests:
cpu: 100m
memory: 256Mi
limits:
cpu: "1"
memory: 2Gi
+22
View File
@@ -0,0 +1,22 @@
# ZFS-backed StorageClass for openebs zfs-localpv (provisioner zfs.csi.openebs.io).
# Standalone resource — NOT part of the openebs Helm release. Backs PVCs on ZFS pool
# `data/ceph`. Apply with: kubectl apply -f storageclasses.yaml
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: localpv-zfs-ceph
provisioner: zfs.csi.openebs.io
reclaimPolicy: Delete
volumeBindingMode: WaitForFirstConsumer
allowVolumeExpansion: true
parameters:
poolname: data/ceph # ZFS dataset the volumes are carved from
csi.storage.k8s.io/fstype: xfs
compression: lz4
dedup: "off"
atime: "off"
relatime: "on"
recordsize: 128k
volblocksize: 4k
thinprovision: "yes"
xattr: sa
+26
View File
@@ -0,0 +1,26 @@
# openebs umbrella chart (openebs/openebs) values — single-node homelab.
# Install/upgrade:
# helm upgrade --install openebs openebs/openebs --version 4.4.0 -n openebs -f values.yaml
# Local ZFS provisioner ON — backs the cluster's PVCs (StorageClass localpv-zfs-ceph on
# pool data/ceph, see storageclasses.yaml). LVM + rawfile OFF (unused — no LVM
# StorageClasses/PVs/volumes). Replicated (Mayastor) OFF. Pinned explicitly so a future
# chart-default change can't silently flip storage engines.
engines:
local:
zfs:
enabled: true
lvm:
enabled: false
rawfile:
enabled: false
replicated:
mayastor:
enabled: false
# The bundled Loki (logging) defaults to 3 SingleBinary replicas with one-pod-per-node
# anti-affinity. On this single node, replicas 2 & 3 are permanently Pending. Filesystem-
# backed SingleBinary Loki should be 1 replica anyway.
loki:
singleBinary:
replicas: 1