feat(acl): keys + ACLs reach driver-only nodes via the artifact (#520 stage 4)

A driver-only node has had no ConfigDb since per-cluster mesh Phase 4, so it
registered NullOpcUaUserAuthenticator and denied every UserName token. Its keys
now ride the deployment artifact instead.

- ConfigComposer snapshots ApiKeys alongside the NodeAcls it already carried.
  Only the peppered HMAC ships, never a secret, and the pepper is node config
  that is deliberately not in the artifact.
- DeploymentArtifact.ParseDataPlaneAuthority decodes both node-side. Grants are
  scoped to the node's own cluster; keys are not, because ApiKey carries no
  ClusterId — a key is a fleet-wide credential and the ACL rows are what bind it.
- OpcUaPublishActor installs them on DataPlaneAuthorityState on every apply,
  BEFORE its plan.IsEmpty early-return: an artifact that only revokes a key
  produces an empty address-space plan, so feeding after that return would leave
  the revoked key working until an unrelated structural change redeployed.
- ArtifactApiKeyStore serves them to session activation; a node with a ConfigDb
  keeps ConfigDbApiKeyStore, whose per-request read is what makes revocation
  immediate there.
- OtOpcUaNodeManager gains DataPlaneAuthority — the cluster identity blocker #4
  named. NodeScope.ClusterId is required and the evaluator fails closed on a
  mismatch, so until now no scope it built could have evaluated. Nothing reads it
  until stages 5/6.

The generation is a node-local COUNTER driven by the RevisionHash, not derived
from its bits. PermissionTrieCache treats the highest GenerationId as current, so
a generation taken from hash bytes would install a lower number about half the
time and the cache would go on serving the previous deployment's ACLs, silently.

Sessions are not stamped with it yet — that belongs with the first seam that
reads a UserAuthorizationState, in stage 5, rather than being built now and
consulted by nothing.

Live-gated on docker-dev against site-a-1 (driver-only, no ConfigDb): key
authenticates, wrong secret denied, both halves of the revocation asymmetry
observed, and a SITE-A grant confirmed not to reach site-b-1. Key minted with an
independent Python HMAC so the verifier was checked against something other than
itself. Rig left clean.

Claude-Session: https://claude.ai/code/session_015p7wGqy3YpZNCpDzTpGMKo
This commit is contained in:
Joseph Doherty
2026-07-31 14:37:18 -04:00
parent c1771dc271
commit a438ac1640
21 changed files with 1505 additions and 48 deletions
@@ -0,0 +1,80 @@
using ZB.MOM.WW.OtOpcUa.Configuration.Entities;
// Fully-qualified rather than relative: this file's own namespace root contains a `Configuration`
// namespace (ZB.MOM.WW.OtOpcUa.Host.Configuration), so a relative `Configuration.Enums` binds there and
// fails to compile.
using ZB.MOM.WW.OtOpcUa.Configuration.Enums;
using ZB.MOM.WW.OtOpcUa.Core.Authorization;
namespace ZB.MOM.WW.OtOpcUa.Host.OpcUa;
/// <summary>
/// <see cref="IApiKeyStore"/> served from the deployment artifact (#520, stage 4). This is the key
/// source for a <b>driver-only</b> node, which has had no ConfigDb since per-cluster mesh Phase 4 and
/// until now denied every UserName token.
/// </summary>
/// <remarks>
/// <para>
/// Purely in-memory: <see cref="OpcUaPublishActor"/> decodes the keys and grants once per applied
/// artifact and installs them on <see cref="IDataPlaneAuthoritySource"/>. That matters because the
/// SDK raises the impersonation callback <b>synchronously, under a <c>SessionManager</c>-wide
/// lock</b> — an I/O-bound lookup here would serialize every session activation in the server
/// behind one disk or network round trip.
/// </para>
/// <para>
/// ⚠️ <b>Consequence: revocation lands on the next deployment, not immediately.</b> A node reading
/// the ConfigDb sees <c>Enabled = 0</c> on the next request; this one sees it only once a new
/// artifact arrives. That asymmetry is inherent to a node with no config database and is
/// documented for operators on the AdminUI key page.
/// </para>
/// <para>
/// Before the first artifact is applied the snapshot is empty, so every key is denied. That is the
/// safe direction, and it is also honest — a node that has not been deployed to has no configured
/// keys.
/// </para>
/// </remarks>
public sealed class ArtifactApiKeyStore : IApiKeyStore
{
private readonly IDataPlaneAuthoritySource _authority;
/// <summary>Initializes a new instance of the <see cref="ArtifactApiKeyStore"/> class.</summary>
/// <param name="authority">The node's current data-plane authorization inputs.</param>
public ArtifactApiKeyStore(IDataPlaneAuthoritySource authority)
{
ArgumentNullException.ThrowIfNull(authority);
_authority = authority;
}
/// <inheritdoc />
public Task<ApiKeyLookup?> FindAsync(string apiKeyId, DateTime utcNow, CancellationToken ct)
{
// Read the snapshot ONCE. Re-reading Current for the grants would let a deploy land between the
// two reads and pair a key from one generation with grants from another.
var snapshot = _authority.Current;
var key = snapshot.FindUsableKey(apiKeyId, utcNow);
return Task.FromResult(key is null
? null
: new ApiKeyLookup(key, ClusterGrants(snapshot, key.Principal)));
}
/// <inheritdoc />
public Task<IReadOnlyList<NodeAcl>> FindGrantsAsync(string principal, CancellationToken ct) =>
Task.FromResult(ClusterGrants(_authority.Current, principal));
/// <summary>
/// Narrows a principal's grants to the Cluster-scoped ones, matching <c>ConfigDbApiKeyStore</c>.
/// Only those confer an interim role (see <see cref="ApiKeyInterimRoleMapper"/>); stage 5 widens
/// both stores to the full grant set once scope is actually enforced.
/// </summary>
private static IReadOnlyList<NodeAcl> ClusterGrants(DataPlaneAuthoritySnapshot snapshot, string principal)
{
var all = snapshot.GrantsFor(principal);
if (all.Count == 0) return [];
var scoped = new List<NodeAcl>(all.Count);
foreach (var grant in all)
{
if (grant.ScopeKind == NodeAclScopeKind.Cluster) scoped.Add(grant);
}
return scoped;
}
}