Files
diskless-workstation/serving/fleet-manager/server-pki.sh
T
Audiobuzz a89c3d2af1 fleet-manager: fix CA install, add register-host, surface agent status
Four fixes from the on-host agent bring-up:

1. Bake the real fleet CA into the image. New `fleet_ca` build var (default =
   the committed placeholder) file-provisions the trust anchor to
   /etc/dot1x/radius-ca.pem — used by wpa_supplicant (RADIUS) AND the fleet-agent
   (verify the manager). CI materializes a FLEET_CA secret to a runner-temp file
   and passes -var fleet_ca; unset => placeholder (lab). No CA material committed.

2. server-pki.sh installs the agent CLIENT CA as root+intermediate (not root
   only). Host TPM certs are leaf-only in NV and signed by the step-ca
   intermediate, so a root-only client CA makes the manager advertise only the
   root in its CertificateRequest -> Go clients send no cert ("tls: certificate
   required") and it couldn't verify a leaf-only cert anyway. bootstrap.sh now
   exports intermediate_ca.crt so FLEET_CA_INTERMEDIATE resolves.

3. New `fleet-manager register-host <serial> [-name] [-hostnqn]` subcommand:
   seeds/updates an inventory record for a host enrolled out of band (bench
   broker, migrated fleet) through the real store, so its agent stream is
   accepted. The manual counterpart to the authorise-approve flow.

4. The agent surfaces the authoritative recv-side gRPC status instead of the
   bare io.EOF gRPC returns on the send side when the server ends the stream —
   so "host is not enrolled" (PermissionDenied) shows up instead of "EOF".
2026-07-10 05:51:57 +00:00

94 lines
4.5 KiB
Bash
Executable File

#!/usr/bin/env bash
# server-pki.sh — issue the fleet-manager AGENT-plane server cert from the fleet
# CA, and stage the fleet root as the agent client-verify CA.
#
# The agent plane is gRPC + mTLS: enrolled hosts present their TPM identity cert
# (fleet CA), and they verify the MANAGER's server cert against the same fleet
# root they hold. So the manager's agent-plane server cert MUST be signed by the
# fleet CA — this script issues it via `step ca certificate` against the running
# fleet step-ca (serving/dot1x-bench/), the same CA that signs enrolment certs.
#
# This does NOT issue:
# - the ADMIN-plane UI cert -> owned by the Traefik ingress (Let's Encrypt
# DNS-01); the manager never terminates TLS. See DEPLOY.md §2c.
# - the manager's BROKER client cert -> ../enrol-broker/admin-pki.sh (admin CA):
# ../enrol-broker/admin-pki.sh ../dot1x-bench/pki/broker fleet-manager
#
# Usage: server-pki.sh
# env overrides:
# OUT=./pki/agent output dir
# AGENT_SANS=diskless-server.bitasylum.org,10.0.0.1 SANs hosts dial (DNS/IP, comma-sep)
# FLEET_CA_URL=https://ca.fleet.lab:8443
# FLEET_CA_ROOT=../dot1x-bench/pki/root_ca.crt
# STEP_PROVISIONER=enrol
# STEP_PROVISIONER_PASSWORD_FILE=../dot1x-bench/pki/stepca/secrets/password
# CERT_NOT_AFTER=8760h
set -euo pipefail
OUT=${OUT:-./pki/agent}
AGENT_SANS=${AGENT_SANS:-diskless-server.bitasylum.org}
FLEET_CA_URL=${FLEET_CA_URL:-https://ca.fleet.lab:8443}
FLEET_CA_ROOT=${FLEET_CA_ROOT:-../dot1x-bench/pki/root_ca.crt}
FLEET_CA_INTERMEDIATE=${FLEET_CA_INTERMEDIATE:-../dot1x-bench/pki/intermediate_ca.crt}
STEP_PROVISIONER=${STEP_PROVISIONER:-enrol}
STEP_PROVISIONER_PASSWORD_FILE=${STEP_PROVISIONER_PASSWORD_FILE:-../dot1x-bench/pki/stepca/secrets/password}
CERT_NOT_AFTER=${CERT_NOT_AFTER:-8760h}
command -v step >/dev/null || { echo "error: 'step' CLI not found (smallstep/cli)"; exit 1; }
[ -f "$FLEET_CA_ROOT" ] || { echo "error: fleet root not found at $FLEET_CA_ROOT (bootstrap the fleet CA first)"; exit 1; }
[ -f "$STEP_PROVISIONER_PASSWORD_FILE" ] || { echo "error: provisioner password not found at $STEP_PROVISIONER_PASSWORD_FILE"; exit 1; }
mkdir -p "$OUT"
# First SAN is the CN; every SAN (DNS name / IP) hosts might dial goes in --san.
primary=${AGENT_SANS%%,*}
san_args=()
IFS=',' read -ra sans <<<"$AGENT_SANS"
for s in "${sans[@]}"; do san_args+=(--san "$s"); done
echo "==> issuing agent-plane server cert (CN=$primary, SANs=$AGENT_SANS)"
step ca certificate "$primary" "$OUT/server.crt" "$OUT/server.key" \
--ca-url "$FLEET_CA_URL" \
--root "$FLEET_CA_ROOT" \
--provisioner "$STEP_PROVISIONER" \
--provisioner-password-file "$STEP_PROVISIONER_PASSWORD_FILE" \
--not-after "$CERT_NOT_AFTER" \
--force \
"${san_args[@]}"
chmod 600 "$OUT/server.key"
# fleet-ca.crt = the agent-plane CLIENT CA that verifies host TPM identity certs.
# Host certs are stored LEAF-ONLY in TPM NV and are signed by the step-ca
# INTERMEDIATE, so this MUST be root + intermediate. Root-only fails two ways:
# (1) the manager can't build leaf->intermediate->root, and (2) it advertises only
# the root in its TLS CertificateRequest, so a Go client whose leaf is issued by
# the intermediate sends NO cert at all ("remote error: tls: certificate
# required"). Prefer an explicit intermediate file; else pull it from the chain
# step wrote into the just-issued server.crt (every cert after the leaf).
intermediate=""
if [ -f "$FLEET_CA_INTERMEDIATE" ]; then
intermediate=$(cat "$FLEET_CA_INTERMEDIATE")
else
intermediate=$(awk 'BEGIN{c=0} /-----BEGIN CERTIFICATE-----/{c++} c>1' "$OUT/server.crt")
fi
{
[ -n "$intermediate" ] && printf '%s\n' "$intermediate"
cat "$FLEET_CA_ROOT"
} > "$OUT/fleet-ca.crt"
if [ -z "$intermediate" ]; then
echo "WARNING: no intermediate CA found (set FLEET_CA_INTERMEDIATE, or the CA has none) —" >&2
echo " fleet-ca.crt is ROOT-ONLY; leaf-only host certs issued by an intermediate" >&2
echo " will fail mTLS with 'tls: certificate required'." >&2
fi
echo
echo "done. staged in $OUT :"
echo " server.crt / server.key -> FLEET_MGR_AGENT_CERT / _KEY"
echo " fleet-ca.crt -> FLEET_MGR_AGENT_CLIENT_CA (verifies host TPM certs)"
echo
echo "next:"
echo " - admin UI cert : owned by the Traefik ingress (LE DNS-01) — DEPLOY.md §2c"
echo " - broker client : ../enrol-broker/admin-pki.sh ../dot1x-bench/pki/broker fleet-manager"
echo " then: cp -r ../dot1x-bench/pki/broker ./pki/broker"