Compare commits

..

1 Commits

Author SHA1 Message Date
Andrew Stoltz
46bbd00d09 Add step-ca agent issuer manifest 2026-05-19 17:52:58 -05:00
6 changed files with 121 additions and 191 deletions

View File

@@ -103,7 +103,6 @@ curl -sk -X DELETE https://dns.iamworkin.lan/api/v1/servers/<serverId>/zones/iam
- **Public read-only hosts**: if a public host fronts a service that also exposes admin writes internally, add a Traefik route match like `Host(...) && (Method(GET) || Method(HEAD))` on the public edge instead of trusting the app to reject unsafe methods. - **Public read-only hosts**: if a public host fronts a service that also exposes admin writes internally, add a Traefik route match like `Host(...) && (Method(GET) || Method(HEAD))` on the public edge instead of trusting the app to reject unsafe methods.
- **Public read-write allowlist hosts**: if a public host accepts a tightly bounded write surface (e.g. bootstrap-JWT POST), pin the allowlist as `(Method(GET) || Method(HEAD) || Method(POST) || Method(OPTIONS))`. PUT/PATCH/DELETE must still 404 at the route. Track A's `updatecenter.iamworkin.lan` / `updates.iamworkin.lan` are the canonical example. The lint test enforces this invariant. - **Public read-write allowlist hosts**: if a public host accepts a tightly bounded write surface (e.g. bootstrap-JWT POST), pin the allowlist as `(Method(GET) || Method(HEAD) || Method(POST) || Method(OPTIONS))`. PUT/PATCH/DELETE must still 404 at the route. Track A's `updatecenter.iamworkin.lan` / `updates.iamworkin.lan` are the canonical example. The lint test enforces this invariant.
- **Traefik VIP netpols**: when a `NetworkPolicy` allows `10.0.56.200`, also allow the post-DNAT backend ports (`8443` for TLS plus `8080` or `8000` for HTTP) or Calico will drop the rewritten flow. - **Traefik VIP netpols**: when a `NetworkPolicy` allows `10.0.56.200`, also allow the post-DNAT backend ports (`8443` for TLS plus `8080` or `8000` for HTTP) or Calico will drop the rewritten flow.
- **RemoteDesktop isolation**: `apps/fc-desktop/network-policies.yaml` intentionally keeps desktop pod egress to named CoreDNS, `intranet-web:5300/TCP`, and noc1 step-ca `10.0.56.10:9000/9443` only. Guacamole display egress is owned separately by `apps/guacamole/guacamole.yaml` through `guacd-desktop-egress` on `5901/TCP`.
- **Auth-safe probes**: services behind API-key or global auth middleware should prefer `tcpSocket` probes unless `/health` is explicitly exempted before the middleware runs. - **Auth-safe probes**: services behind API-key or global auth middleware should prefer `tcpSocket` probes unless `/health` is explicitly exempted before the middleware runs.
- **ArgoCD must use internal Gitea URL**: `http://gitea-clusterip.gitea.svc.cluster.local:3000/bluejay/bluejay-infra.git`, not the external HTTPS URL (step-ca cert isn't trusted by ArgoCD). The `ApplicationSet` and any hand-created `Application` must both use the internal URL. - **ArgoCD must use internal Gitea URL**: `http://gitea-clusterip.gitea.svc.cluster.local:3000/bluejay/bluejay-infra.git`, not the external HTTPS URL (step-ca cert isn't trusted by ArgoCD). The `ApplicationSet` and any hand-created `Application` must both use the internal URL.

View File

@@ -20,12 +20,9 @@
# 1) desktop-isolation — Browser Lab session pods. # 1) desktop-isolation — Browser Lab session pods.
# #
# Locks down pods labeled `app.kubernetes.io/name=remote-desktop` (every # Locks down pods labeled `app.kubernetes.io/name=remote-desktop` (every
# session pod regardless of template). Allows guacd ingress for the display # session pod regardless of template). Allows guacd ingress for the VNC/RDP
# lane and remotedesktop-web's pre-handoff probing. Egress is deliberately # display lane and remotedesktop-web's pre-handoff probing. Egress: NFS to
# narrow: named CoreDNS, direct Intranet web, and noc1 step-ca only. There is # Synology, DNS, Traefik (cluster + LB VIP), Intranet (Browser Lab home).
# no broad Traefik/VIP or internet egress from desktop sessions. If a future
# Browser Lab path needs a public-style host, prefer an explicit Service rule
# or include the post-DNAT backend port per the Traefik VIP lint.
apiVersion: networking.k8s.io/v1 apiVersion: networking.k8s.io/v1
kind: NetworkPolicy kind: NetworkPolicy
metadata: metadata:
@@ -68,22 +65,51 @@ spec:
- port: 5901 - port: 5901
protocol: TCP protocol: TCP
egress: egress:
# CoreDNS only. The old to: [] DNS rule accidentally allowed any DNS # NFS to Synology
# listener in any namespace or routed network.
- to: - to:
- namespaceSelector: - ipBlock:
matchLabels: cidr: 10.0.58.3/32
kubernetes.io/metadata.name: kube-system ports:
podSelector: - port: 2049
matchLabels: protocol: TCP
k8s-app: kube-dns - port: 2049
protocol: UDP
- port: 111
protocol: TCP
- port: 111
protocol: UDP
- to:
- ipBlock:
cidr: 10.0.58.3/32
ports:
- port: 445
protocol: TCP
- to: []
ports: ports:
- port: 53 - port: 53
protocol: UDP protocol: UDP
- port: 53 - port: 53
protocol: TCP protocol: TCP
# Browser Lab home / internal docs target. Use the real service port - to:
# directly rather than public Traefik host aliases. - ipBlock:
cidr: 10.0.56.200/32
- ipBlock:
cidr: 10.43.33.87/32
- namespaceSelector:
matchLabels:
kubernetes.io/metadata.name: traefik-system
podSelector:
matchLabels:
app.kubernetes.io/name: traefik
ports:
- port: 80
protocol: TCP
- port: 443
protocol: TCP
- port: 8000
protocol: TCP
- port: 8443
protocol: TCP
- to: - to:
- namespaceSelector: - namespaceSelector:
matchLabels: matchLabels:
@@ -94,17 +120,6 @@ spec:
ports: ports:
- port: 5300 - port: 5300
protocol: TCP protocol: TCP
# noc1 step-ca ACME endpoint. The lane brief called out 9000/TCP; the live
# ACME directory currently answers on 9443/TCP, so both stay pinned to the
# same host rather than reopening Traefik or internet egress.
- to:
- ipBlock:
cidr: 10.0.56.10/32
ports:
- port: 9000
protocol: TCP
- port: 9443
protocol: TCP
--- ---
# 2) fc-desktop-default-deny — namespace-wide catch-all. # 2) fc-desktop-default-deny — namespace-wide catch-all.
# #
@@ -315,11 +330,3 @@ spec:
protocol: UDP protocol: UDP
- port: 53 - port: 53
protocol: TCP protocol: TCP
- to:
- ipBlock:
cidr: 10.0.56.10/32
ports:
- port: 9000
protocol: TCP
- port: 9443
protocol: TCP

View File

@@ -0,0 +1,31 @@
# Step issuer for FlowerCore.DeviceManagement runtime mTLS leaves.
#
# Requires the smallstep step-issuer CRDs/controller:
# stepclusterissuers.certmanager.step.sm
# The provisioner password lives in the live cert-manager Secret below; do not
# commit the password or generated private key material to this repo.
apiVersion: certmanager.step.sm/v1beta1
kind: StepClusterIssuer
metadata:
name: step-ca-agent
labels:
app.kubernetes.io/name: step-ca-agent
app.kubernetes.io/component: pki
app.kubernetes.io/part-of: flowercore
app.kubernetes.io/managed-by: argocd
flowercore.io/tenant-id: system
flowercore.io/created-by: bluejay-infra
annotations:
flowercore.io/traceability-standard: k8s-pod-ownership-and-traceability-standard
flowercore.io/provisioner-source: profile::pki::stepca
flowercore.io/secret-source: cert-manager/step-ca-agent-provisioner-password
spec:
url: https://10.0.56.10:9443
caBundle: LS0tLS1CRUdJTiBDRVJUSUZJQ0FURS0tLS0tCk1JSUJ4RENDQVdxZ0F3SUJBZ0lSQVBZMzU3RzZvdzZ6TUFMNSs0YlMya2t3Q2dZSUtvWkl6ajBFQXdJd1FERWEKTUJnR0ExVUVDaE1SU1VGdFYyOXlhMmx1SUVGRFRVVWdRMEV4SWpBZ0JnTlZCQU1UR1VsQmJWZHZjbXRwYmlCQgpRMDFGSUVOQklGSnZiM1FnUTBFd0hoY05Nall3TXpBNE1UZ3dOekV4V2hjTk16WXdNekExTVRnd056RXhXakJBCk1Sb3dHQVlEVlFRS0V4RkpRVzFYYjNKcmFXNGdRVU5OUlNCRFFURWlNQ0FHQTFVRUF4TVpTVUZ0VjI5eWEybHUKSUVGRFRVVWdRMEVnVW05dmRDQkRRVEJaTUJNR0J5cUdTTTQ5QWdFR0NDcUdTTTQ5QXdFSEEwSUFCSjJuMDRYMQpKWm81WmRxL2kxSWR2OCtmcXdaeUF6Qmg3d2hicWowU1dzSkw4VVdSYWJDTXFZQ3M3K2RYTzB4UlN6cWt3RkRMCngrdm9vT2FpOFJnUk5oYWpSVEJETUE0R0ExVWREd0VCL3dRRUF3SUJCakFTQmdOVkhSTUJBZjhFQ0RBR0FRSC8KQWdFQk1CMEdBMVVkRGdRV0JCUm51UFBRUjZpTS9INnZPbHVpVTNTeWdheXo4akFLQmdncWhrak9QUVFEQWdOSQpBREJGQWlFQXJRSzlkWVBHbUFac2RZbmp6aXVGVlZFNU5LWlVjY2VZdkdmR0MrdExYVXNDSUF1ZEYyekpyQ1JxCjNtSzUwWlpFVC9md1RrSndpRUY0ODI0bWpQOHAxQ0tNCi0tLS0tRU5EIENFUlRJRklDQVRFLS0tLS0K
provisioner:
name: step-ca-agent
kid: RF3A9welUYVOWBX8tr19aWyA2kQlxoGZN1dRwTElUEM
passwordRef:
name: step-ca-agent-provisioner-password
namespace: cert-manager
key: password

View File

@@ -254,68 +254,6 @@ spec:
targetPort: 4822 targetPort: 4822
name: guacd name: guacd
--- ---
# Guacd display egress isolation.
#
# Guacamole web talks to guacd on TCP/4822. Guacd then opens the desktop
# display connection to the per-session pod. Keep that second hop at raw VNC
# 5901/TCP for the current RemoteDesktop Browser Lab/openSUSE images. Do not
# grant guacd broad fc-desktop namespace egress; desktop-to-desktop lateral
# paths remain blocked by apps/fc-desktop/network-policies.yaml.
apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
name: guacd-desktop-egress
namespace: guacamole
labels:
app.kubernetes.io/part-of: remotedesktop
app.kubernetes.io/component: display-isolation
spec:
podSelector:
matchLabels:
app: guacd
policyTypes:
- Ingress
- Egress
ingress:
- from:
- podSelector:
matchLabels:
app: guacamole
ports:
- port: 4822
protocol: TCP
egress:
- to:
- namespaceSelector:
matchLabels:
kubernetes.io/metadata.name: kube-system
podSelector:
matchLabels:
k8s-app: kube-dns
ports:
- port: 53
protocol: UDP
- port: 53
protocol: TCP
# kubectl-proxy sidecar reaches the Kubernetes API; keep it explicit
# because this NetworkPolicy selects the whole guacd pod.
- to: []
ports:
- port: 443
protocol: TCP
- port: 6443
protocol: TCP
- to:
- namespaceSelector:
matchLabels:
kubernetes.io/metadata.name: fc-desktop
podSelector:
matchLabels:
app.kubernetes.io/name: remote-desktop
ports:
- port: 5901
protocol: TCP
---
# Guacamole Web Application # Guacamole Web Application
apiVersion: apps/v1 apiVersion: apps/v1
kind: Deployment kind: Deployment

View File

@@ -432,6 +432,7 @@ public sealed class FleetManifestLintTests
"1password-item.yaml", "1password-item.yaml",
"argocd-application.yaml", "argocd-application.yaml",
"certificate-web.yaml", "certificate-web.yaml",
"clusterissuer-step-ca-agent.yaml",
"clusterrole-operator.yaml", "clusterrole-operator.yaml",
"clusterrolebinding-operator.yaml", "clusterrolebinding-operator.yaml",
"deployment-operator.yaml", "deployment-operator.yaml",
@@ -516,6 +517,53 @@ public sealed class FleetManifestLintTests
.ContainSingle("devices.iamworkin.lan"); .ContainSingle("devices.iamworkin.lan");
} }
[Fact]
public void FcDeviceManagement_StepCaAgentIssuerMustTargetNocProvisioner()
{
var issuer = FcDeviceManagementDocuments()
.Single(document => document.Kind == "StepClusterIssuer" && document.Name == "step-ca-agent");
issuer.Scalar("apiVersion").Should().Be("certmanager.step.sm/v1beta1");
issuer.Scalar("spec", "url").Should().Be("https://10.0.56.10:9443");
issuer.Scalar("spec", "caBundle").Should().NotBeNullOrWhiteSpace();
issuer.Scalar("spec", "provisioner", "name").Should().Be("step-ca-agent");
issuer.Scalar("spec", "provisioner", "kid").Should().Be("RF3A9welUYVOWBX8tr19aWyA2kQlxoGZN1dRwTElUEM");
}
[Fact]
public void FcDeviceManagement_StepCaAgentIssuerMustReferencePasswordSecretOnly()
{
var issuer = FcDeviceManagementDocuments()
.Single(document => document.Kind == "StepClusterIssuer" && document.Name == "step-ca-agent");
issuer.Scalar("spec", "provisioner", "passwordRef", "name")
.Should()
.Be("step-ca-agent-provisioner-password");
issuer.Scalar("spec", "provisioner", "passwordRef", "namespace").Should().Be("cert-manager");
issuer.Scalar("spec", "provisioner", "passwordRef", "key").Should().Be("password");
var issuerText = File.ReadAllText(Path.Combine(Inventory.BluejayRoot, "apps", "fc-devicemgmt", "clusterissuer-step-ca-agent.yaml"));
issuerText.Should().NotContain("stringData:");
issuerText.Should().NotContain("password:");
issuerText.Should().NotContain("privateKey");
}
[Fact]
public void FcDeviceManagement_StepCaAgentIssuerMustCarryTraceabilityMetadata()
{
var issuer = FcDeviceManagementDocuments()
.Single(document => document.Kind == "StepClusterIssuer" && document.Name == "step-ca-agent");
issuer.Scalar("metadata", "labels", "app.kubernetes.io/managed-by").Should().Be("argocd");
issuer.Scalar("metadata", "labels", "flowercore.io/tenant-id").Should().Be("system");
issuer.Scalar("metadata", "annotations", "flowercore.io/provisioner-source")
.Should()
.Be("profile::pki::stepca");
issuer.Scalar("metadata", "annotations", "flowercore.io/secret-source")
.Should()
.Be("cert-manager/step-ca-agent-provisioner-password");
}
[Fact] [Fact]
public void FcDeviceManagement_OperatorRbacMustCoverDevicesAndOwnerLookup() public void FcDeviceManagement_OperatorRbacMustCoverDevicesAndOwnerLookup()
{ {

View File

@@ -1,93 +0,0 @@
using FluentAssertions;
using Xunit;
namespace BluejayInfraLint.Tests;
[Trait("Category", "Unit")]
public sealed class RemoteDesktopNetworkPolicyTests
{
private static readonly ManifestInventory Inventory = ManifestInventory.Load();
[Fact]
public void LiveDesktopIsolation_AllowsOnlyCoreDnsIntranetAndStepCaEgress()
{
var policy = NetworkPolicy("fc-desktop", "desktop-isolation");
var ports = policy.EgressPorts().ToHashSet(StringComparer.Ordinal);
ports.Should().BeEquivalentTo("53", "5300", "9000", "9443");
policy.AllScalars().Should().Contain(new[]
{
"kube-system",
"kube-dns",
"intranet",
"intranet-web",
"10.0.56.10/32"
});
}
[Fact]
public void LiveDesktopIsolation_RemovesInternetNfsAndTraefikEgress()
{
var policy = NetworkPolicy("fc-desktop", "desktop-isolation");
var scalars = policy.AllScalars().ToList();
var ports = policy.EgressPorts().ToHashSet(StringComparer.Ordinal);
scalars.Should().NotContain(new[] { "10.0.58.3/32", "10.0.56.200/32", "10.43.33.87/32", "traefik-system" });
ports.Should().NotContain(new[] { "80", "443", "445", "111", "2049", "8000", "8080", "8443" });
policy.MappingSequence("spec", "egress")
.Should()
.NotContain(rule => EgressRuleHasEmptyTo(rule), "desktop sessions must not use to: [] internet-style egress");
}
[Fact]
public void LiveGuacdIsolation_AllowsRawVncToDesktopPodsOnly()
{
var policy = NetworkPolicy("guacamole", "guacd-desktop-egress");
var scalars = policy.AllScalars().ToList();
var ports = policy.EgressPorts().ToHashSet(StringComparer.Ordinal);
ports.Should().Contain("5901");
scalars.Should().Contain(new[] { "fc-desktop", "remote-desktop" });
ports.Should().NotContain(new[] { "3000", "3001", "3389", "80", "8080", "8443" });
}
[Fact]
public void LiveGuacdIsolation_KeepsGuacamoleWebIngressOnGuacdPort()
{
var policy = NetworkPolicy("guacamole", "guacd-desktop-egress");
policy.Scalar("spec", "podSelector", "matchLabels", "app").Should().Be("guacd");
policy.AllScalars().Should().Contain(new[] { "guacamole", "4822" });
}
[Fact]
public void HelperSmoke_FindsExpectedRemoteDesktopPolicies()
{
NetworkPolicy("fc-desktop", "desktop-isolation").Name.Should().Be("desktop-isolation");
NetworkPolicy("guacamole", "guacd-desktop-egress").Name.Should().Be("guacd-desktop-egress");
}
[Fact]
public void HelperSmoke_EgressPortExtractionKeepsDistinctPorts()
{
var ports = NetworkPolicy("fc-desktop", "desktop-isolation")
.EgressPorts()
.ToHashSet(StringComparer.Ordinal);
ports.Should().HaveCount(4);
ports.Should().Contain(new[] { "53", "5300", "9000", "9443" });
}
private static ManifestDocument NetworkPolicy(string ns, string name)
=> Inventory.Documents.Single(document =>
document.Kind == "NetworkPolicy"
&& string.Equals(document.Namespace, ns, StringComparison.Ordinal)
&& string.Equals(document.Name, name, StringComparison.Ordinal));
private static bool EgressRuleHasEmptyTo(YamlDotNet.RepresentationModel.YamlMappingNode rule)
=> rule.Children.Any(entry =>
entry.Key is YamlDotNet.RepresentationModel.YamlScalarNode key
&& string.Equals(key.Value, "to", StringComparison.Ordinal)
&& entry.Value is YamlDotNet.RepresentationModel.YamlSequenceNode sequence
&& sequence.Children.Count == 0);
}