10 Commits
Author SHA1 Message Date
Marc Schier 15eb55c46e Cert rotation: force renegotiate (#3562) + auto-detect on client (#3160) (#3849)
# Description

Implements the OPC UA Part 12 §7.10.9 contract on the server (fixes
#3562) AND the client-side counterpart for immediate detection of
own-cert / trust-list / CRL changes (fixes #3160).

## Server side (#3562)

Per Part 12 §7.10.9 the server _shall_ "force existing SecureChannels
affected by the changes to renegotiate" once the `ApplyChanges` method
response has been delivered. Previously the server reloaded the
certificate but did not cut affected channels; clients kept using the
old cert until natural rekey, and the code carried an explicit `TODO`
next to a 1-second `Task.Delay` hack in
`ConfigurationNodeManager.ApplyChanges`.

Key server changes:

* **New optional capability interface**
`Opc.Ua.ITransportListenerCertificateRotation` (single method
`CloseChannelsForCertificate(Certificate)`) — implemented by
`TcpTransportListener` and `HttpsTransportListener`. Custom transport
listeners can opt in without breaking `ITransportListener`.
* **TCP path**: new internal
`TcpListenerChannel.TryCloseForCertificateRotation(oldThumbprint, out
globalChannelId)` matches the channel's negotiated `ServerCertificate`
by thumbprint, sends `BadCertificateInvalid`, and calls
`ChannelFaulted()` to actually close the socket.
`TcpTransportListener.CloseChannelsForCertificate` snapshots
`m_channels` outside `m_lock` to avoid lock inversion, invokes the
per-channel helper, returns cut global channel ids. **The listener
socket stays bound** — clients can reconnect immediately.
* **HTTPS path**: HTTPS owns the cert at the Kestrel/HTTP.sys listener
level, so a `Stop()`+`Start()` cycle is the only practical
force-renegotiate. Moved out of `CertificateUpdate` (which now does only
the soft fan-out) into `CloseChannelsForCertificate` so the staging
event no longer disrupts in-flight requests prematurely.
* **`ConfigurationNodeManager.ApplyChanges`** rewritten:
- Captures the per-group pre-transaction certificate (preserved across
multi-stage transactions per Part 12 §7.10.2) and the staged
`UpdateCertificateData`.
- Schedules a deferred apply: waits for the response to flush, re-syncs
the registry via `CertificateManager.UpdateAsync`, then calls
`CloseChannelsForCertificate` on every listener that implements the
capability.
- **Configurable grace period**: new
`IConfigurationNodeManager.ApplyChangesGracePeriod` (default 250 ms,
negative clamped to zero). Hosts running over higher-latency
administrative links can raise it.
- Exposes
`IConfigurationNodeManager.DrainPendingApplyChangesAsync(CancellationToken)`
for tests/hosts to deterministically await rotation.
  - Removes the explicit 1-second `Task.Delay` TODO hack.
* **`Opc.Ua.Server.ITransportListenerRegistryProvider`** — narrow opt-in
interface on `ServerInternalData` exposing the live transport listener
list. Lets `ConfigurationNodeManager` drive the fan-out without
expanding `IServerInternal` (binary compat for mocked test fixtures).
* **Sessions are NOT closed**. Per spec, the cut is on the SecureChannel
only; existing `Session.IsSecureChannelValid` + `Session.Activate`
rebind handles transfer over the new channel.

## Client side (#3160)

The client SDK previously had no automatic detection of changes to its
own certificate, trust list, or CRL — users had to call
`Session.ReloadInstanceCertificateAsync` manually and trigger a
reconnect. The spec-mandated `ApplicationCertificateUpdated`,
`TrustListUpdated`, and `CrlUpdated` enum values existed but only the
first was being emitted by `CertificateManager`.

Key client/shared changes:

* **`CertificateManager` now emits `TrustListUpdated` and `CrlUpdated`**
from `WriteTrustListAsync` (per mask) and from
`TrustListTransaction.CommitAsync` (after the atomic apply). The new
internal `NotifyTrustListChanged` also invalidates cached
peer/user/HTTPS validation cores so subsequent validations pick up the
fresh trust state.
* **`ManagedSession.AutoReconnectOnCertificateChange`** opt-in property
(default `false` for backwards compat). When `true`, `ManagedSession`
subscribes to `CertificateManager.CertificateChanges` on construction
and, on the three relevant kinds:
- Calls `Session.ReloadInstanceCertificateAsync` on
`ApplicationCertificateUpdated` so the next ActivateSession is signed
with the rotated client cert.
- Triggers a reconnect via the state machine so the new state is
honoured within milliseconds rather than at the next
`SecurityTokenLifetime` rekey.
* **`ManagedSession.ApplicationCertificateChanged`** event surfaces
every observed change regardless of the auto-reconnect policy for
diagnostics or custom rotation logic.
* Subscription is disposed by the existing `Dispose` / `DisposeAsync`
paths.

```csharp
using var managed = await ManagedSession.CreateAsync(
    configuration, endpoint, sessionFactory);
managed.AutoReconnectOnCertificateChange = true;

managed.ApplicationCertificateChanged += (sender, e) =>
{
    Console.WriteLine($"observed {e.Kind} on {e.TrustList}");
};
```

## Tests

| Suite | net10.0 | net48 |
|---|---|---|
| `TransportListenerCertificateRotationTests` (new, 6 tests) | ✅ | ✅ |
| `CertificateManagerTests` (4 new event tests) | ✅ 22/22 | ✅ 4/4 |
| `PushCertManagement` | 50/53 ✅ | 50/53 ✅ |
| `GDS PushTest` (focused subset) | 15/15 ✅ | 16/16 ✅ |
| `Opc.Ua.Server.Tests` (focused subset) | 38/38 ✅ | 38/38 ✅ |
| `GDS PushTest` (full) | 313/338 ✅ | — |

Skips are pre-existing (cert-type mismatches, optional methods) — no
regressions.

New integration test
`PushTest.ApplyChangesForcesChannelRenegotiateAsync` positively asserts
the stale channel fails with a recognized channel-invalidation status
after `ApplyChanges`, and that a fresh reconnect succeeds with the
rotated cert.

## Documentation

- `Docs/CertificateManager.md` — new sections:
- "Server-Side Certificate Rotation via Push (OPC UA Part 12 §7.10.9)" —
`ITransportListenerCertificateRotation` contract + custom listener
example.
- "Client-Side Auto-Detection of Certificate Changes (issue #3160)" —
`AutoReconnectOnCertificateChange` + `ApplicationCertificateChanged`
event + the three covered scenarios.

## Related Issues

- Fixes #3562
- Fixes #3160

## Checklist

- [ ] I have signed the
[CLA](https://opcfoundation.org/license/cla/ContributorLicenseAgreementv1.0.pdf)
and read the
[CONTRIBUTING](https://github.com/OPCFoundation/UA-.NETStandard/blob/master/CONTRIBUTING.md)
doc.
- [x] I have added tests that prove my fix is effective or that my
feature works and increased code coverage.
- [x] I have added all necessary documentation.
- [x] I have verified that my changes do not introduce (new) build or
analyzer warnings.
- [x] I ran focused tests locally using the **UA.slnx** projects against
.net **framework 4.8** and .net **10**, and all passed. _(Full UA.slnx
test run delegated to CI.)_
- [ ] I fixed **all** failing and flaky tests in the CI pipelines and
**all** CodeQL warnings.
- [ ] I have addressed **all** PR feedback received.
2026-06-08 12:19:49 +02:00
Marc Schier 0000513cde Pluggable user-identity provider model (#3807)
# Description

Introduces a pluggable user-identity provider model covering all OPC UA
Part 6 §6.5 mechanisms (Anonymous, UserName, X509, IssuedToken/JWT) on
both client and server, with full
`Microsoft.Extensions.DependencyInjection` integration, plus the Part 12
§9 `AuthorizationService` two-phase token-request flow **including
refresh-token rotation**, the Part 12 §8 KeyCredential Push model, and a
clean provider model for OPC 10000-12 §7.2 `ApplicationSelfAdmin`. The
work is split across phase-by-phase commits so it can be reviewed
independently; nothing here changes wire-level behaviour or forces
existing applications to migrate.

### Documentation

All under `Docs/`:

- `IdentityProviders.md` — full developer guide: three-layer model,
recipes, configuration reference, "what gets auto-wired" matrix, JWS
algorithm table, identity-augmenters section, **"Implementing your own
provider"** how-to subsections for Entra / OIDC / Windows / AspNetCore
(sibling files merged in per review feedback).
- `DependencyInjection.md` — `Identity` sections with `appsettings.json`
examples, identity-extension matrix, `WithAuthorizationService` +
`WithKeyCredentialPush` + `AddIdentityAugmenter` extensions documented.
- `RoleBasedUserManagement.md` — claim-criteria callout describing
`IIdentityClaims` probe surface.
- `AuthorizationService.md` — Architecture diagram +
`IAccessTokenProvider` conceptual narrative restored from master; modern
`StartRequestToken` / `FinishRequestToken` flow leads; refresh-token
recipe added.
- `KeyCredentialService.md` — Push model + Pull vs Push vs Bridge
decision matrix + Hybrid (Push + Pull) section; explicit experimental
warning for the bridge.
- `GDS.md` — KeyCredential Push, `StartRequestToken`,
`GdsApplicationSelfAdminProvider`, **shipped Refresh-token support with
rotation policy and audit semantics** all documented.
- `MigrationGuide.md` — User Identity Providers section listing
`[Obsolete]` markers + mechanical recipes; SelfAdmin migration via
`IIdentityAugmenter`; `EnableRefreshTokens` default note.
- `README.md` index entries.

## Checklist

- [x] I have signed the
[CLA](https://opcfoundation.org/license/cla/ContributorLicenseAgreementv1.0.pdf)
and read the
[CONTRIBUTING](https://github.com/OPCFoundation/UA-.NETStandard/blob/master/CONTRIBUTING.md)
doc.
- [x] I have added tests that prove my fix is effective or that my
feature works and increased code coverage.
- [x] I have added all necessary documentation.
- [x] I have verified that my changes do not introduce (new) build or
analyzer warnings.
- [x] I ran **all** tests locally using the **UA.slnx** solution against
at least .net **framework** and .net **10**, and all passed.
- [x] I fixed **all** failing and flaky tests in the CI pipelines and
**all** CodeQL warnings.
- [x] I have addressed **all** PR feedback received.
2026-06-02 01:14:54 +02:00
romanett e49d17d706 [Server] Make MonitoredNode aware of IAsyncNodeManager and introduce channel-based queueing under load (#3804)
## Proposed changes

This PR updates the server monitored-item pipeline to be
`IAsyncNodeManager`-aware and introduces channel-based FIFO queueing in
the monitored-node path to improve behavior under load.

It also includes follow-up fixes from review/CI feedback:
- Adjusted
`AsyncCustomNodeManagerTests.WriteAsync_PublishesValueToMonitoredItemQueueAsync`
to account for channel-based delivery timing/coalescing while still
asserting the written value is observed.
- Merged latest `master` into this branch and resolved merge conflicts
(notably in `ConfigurationNodeManager`), keeping async address-space
lookup behavior intact.
- Applied small compile-safety corrections needed for successful
validation.
2026-05-27 10:12:17 +02:00
Marc Schier 51bcd4272a GDS Part 12 Full Compliance (#3795)
* GDS: Authorization & SecureChannel hardening (Part 12 §6.5, §7.2, §7.6)

* Fix OnGetCertificates self-admin authorization: pass applicationId to
  HasAuthorization so ApplicationSelfAdmin can read its own certificates per
  OPC 10000-12 §7.6.10 (was rejecting self-admin pull-management callers).
* Fix OnUnregisterApplicationAsync self-admin authorization: also pass
  applicationId (was silently failing the self-admin branch).
* OnUpdateApplication: accept ApplicationSelfAdmin and ApplicationAdmin in
  addition to DiscoveryAdmin per §6.5.7. Add self-admin role-permission shim.
* OnGetCertificateStatus: accept CertificateAuthorityAdmin / SelfAdmin /
  ApplicationAdmin per §7.6.12 (was using AuthenticatedUserOrSelfAdmin which
  is more permissive than spec).

* Introduce ApplicationAdmin privilege (Part 12 §7.2):
  - Add GdsRole.ApplicationAdmin static privilege.
  - Extend GdsRoleBasedIdentity with AdministeredApplicationIds and a new
    constructor accepting the list. WithAdditionalRoles preserves it.
  - AuthorizationHelper.HasAuthorization recognises the privilege: alone
    when no applicationId is supplied (e.g. RegisterApplication) and
    membership-tested against AdministeredApplicationIds otherwise.
  - HasTrustListAccess honours the ApplicationAdmin privilege for the
    administered applications' trust lists.
  - New role lists: DiscoveryAdminOrAppAdmin (RegisterApplication),
    DiscoveryAdminOrSelfAdminOrAppAdmin (Update/UnregisterApplication),
    CertificateAuthorityAdminOrSelfAdminOrAppAdmin (cert management).

* SecureChannel result-code alignment (Part 12 method tables):
  - HasAuthenticatedSecureChannel throws BadSecurityModeInsufficient (was
    BadUserAccessDenied).
  - Adds requireEncryption flag so callers can demand SignAndEncrypt
    (used by future StartNewKeyPairRequest/FinishRequest tightening).
  - Sign and SignAndEncrypt are now both accepted as 'authenticated'.

* Tests:
  - Register the GDS namespace in AuthorizationHelperTests SetUp so
    GDS-namespace role NodeIds do not collide on NodeId.Null with
    privilege roles.
  - Update HasAuthenticatedSecureChannelThrowsForNonSecureChannel to
    expect BadSecurityModeInsufficient.
  - Add HasAuthenticatedSecureChannelDoesNotThrowForSign and
    HasAuthenticatedSecureChannelWithRequireEncryptionThrowsForSign.
  - Add ApplicationAdmin coverage: SucceedsForAdministeredApp,
    ThrowsForUnmanagedApp, SucceedsForRegisterApplication,
    ThrowsWithEmptyAdministeredList.
  - Add NewRoleListsAreCorrectlyPopulated regression test.

The framework already enforces SigningRequired/EncryptionRequired via
DefaultAccessRestrictions in OpcUaGdsModel.xml, returning
BadSecurityModeInsufficient at request dispatch time
(MasterNodeManager.ValidateAccessRestrictions). The model declares the
correct restriction on every admin method, so explicit handler-level
checks would be redundant. Verified manually against §6.5.6, §6.5.7,
§6.5.8, §7.6.4-7.6.10 and §7.6.12.

23 Authorization unit tests pass. 181 unit + 420 integration GDS tests
pass on net10.0; build clean on net48 and net10.0.

* GDS: Audit event correctness (Part 12 7.6, 7.9)

* Fix audit-event ordering: emit CertificateRequestedAuditEvent AFTER the
  method outcome is known. Previously the event was raised before
  authorization with Status=true, so any failed call appeared in audit as
  a successful certificate request and any authorization-denied call
  generated a misleading 'success' audit trail.
  - OnStartNewKeyPairRequest: refactor to try/catch/finally, emit audit
    once with auditException reflecting either the captured throw, the
    bad ServiceResult, or null on success.
  - OnStartSigningRequestAsync: add the audit emission (previously
    missing entirely - 7.9.3 requires it) using the same pattern.
* Redact privateKeyPassword from audit InputArguments. The previous
  ApplicationsNodeManager.cs:1206-1223 included the raw password string
  as a Variant in the audit event payload, leaking the secret into any
  downstream audit sink. The password is now replaced with
  AuditEvents.RedactedPrivateKeyPassword (placeholder string).
* OnFinishRequestAsync: build the audit InputArguments from only the
  actual method inputs (applicationId, requestId) per 7.6.6 method
  signature. Previously the implementation appended the returned
  Certificate / PrivateKey / IssuerCertificates to InputArguments, which
  leaked the returned PrivateKey into audit payloads.
* Raise CertificateRevokedAuditEventType in OnRevokeCertificate on
  success and failure per 7.6.9. The event type was defined in the
  model (OpcUaGdsModel.xml:543) but never emitted. Refactored
  OnRevokeCertificate with try/catch/finally; sync-over-async (.Result)
  is preserved here and addressed separately in Track C (c-revoke-async).

* AuditEvents helper:
  - New ReportCertificateRevokedAuditEvent extension method, mirroring
    the existing CertificateRequested/Delivered helpers, with an
    Exception parameter that drives Status (true on success, false on
    failure).
  - New RedactedPrivateKeyPassword and RedactedPrivateKey constants
    documented as the canonical placeholders. Class-level remarks
    instruct callers to redact sensitive arguments before invoking the
    helpers.

* Tests:
  - AuditEventsTests: lock the placeholder values
    (RedactedPrivateKeyPassword and RedactedPrivateKey) so they remain
    stable for downstream audit-sink filters.
  - Existing 181 unit + 420 integration GDS tests pass on net10.0; build
    clean on net48 and net10.0.

b-keycred-audit and b-token-audit remain pending pending Track E and F
respectively (KeyCredentialService and AuthorizationService handlers).

* GDS: Directory / Pull Management correctness gaps (Part 12 6.5, 7.6)

* OnRevokeCertificate converted to async OnRevokeCertificateAsync; uses
  await instead of sync-over-async .Result, accepts the method-call
  cancellation token, and is now wired through OnCallAsync to integrate
  with the async dispatch path. RevokeCertificateAsync helper takes an
  optional CancellationToken so callers (OnUnregisterApplicationAsync,
  OnRevokeCertificateAsync) can forward it.

* OnCheckRevocationStatusAsync now computes the ValidityTime per
  OPC 10000-12 7.6.11. The previous implementation always returned
  DateTime.MinValue. The new implementation enumerates the CRLs in the
  trusted-issuer store and uses the earliest NextUpdate as ValidityTime
  when the chain validates. On Bad results (chain validation failed)
  ValidityTime remains DateTime.MinValue so callers re-check
  immediately.

* OnFinishRequestAsync returns the issuer chain via a new
  BuildIssuerCertificateChain helper. The previous implementation
  returned only the immediate signing CA, which is incomplete for
  multi-tier CAs per OPC 10000-12 7.6.6. The helper uses
  ICertificateGroup.Certificates (in-memory) as the X509Chain
  ExtraStore and falls back to the immediate issuing CA if chain
  building does not yield a chain. The helper is intentionally
  synchronous to avoid contending with concurrent SigningRequestAsync /
  NewKeyPairRequestAsync writes against the CertificateGroup's
  AuthoritiesStore (which caused parameterised PushTest cases to fail
  in parallel runs).

* ValidateApplication (ApplicationsDatabaseBase) now enforces the
  reverse-connect rules in OPC 10000-12 6.5.5:
  - Server / DiscoveryServer DiscoveryUrls must NOT start with the
    "rcp+" prefix.
  - Client DiscoveryUrls (when non-empty) must all start with "rcp+"
    AND ServerCapabilities must include "RCP".
  - ClientAndServer must register at least one non-rcp+ Server
    DiscoveryUrl; reverse-connect URLs are still permitted alongside
    and require the RCP ServerCapability.
  Removed the previous blanket rejection of any DiscoveryUrls on
  ApplicationType.Client (which contradicted the spec's reverse-connect
  registration pattern).

* Tests:
  - RegisterApplicationClientWithRcpDiscoveryUrlsAndRcpCapabilityDoesNotThrow
  - RegisterApplicationClientWithRcpDiscoveryUrlsAndMissingCapabilityThrows
  - RegisterApplicationServerWithRcpDiscoveryUrlThrows
  - RegisterApplicationClientAndServerWithMixedUrlsDoesNotThrow
  - RegisterApplicationClientAndServerWithOnlyRcpDiscoveryUrlsThrows
  All 188 unit + 420 integration GDS tests pass on net10.0; build clean
  on net48 and net10.0.

c-redundancy intentionally requires no enforcement: both transparent
(single application with multiple DiscoveryUrls) and non-transparent
(separate applications with NTRS capability) patterns per OPC 10000-12
6.5.6 are admitted by the validator; transparent redundancy is a
registrant convention rather than a server-enforceable invariant.


* Annex D capabilities: Self-advertise the LDS-ME capability identifier
  whenever the MulticastExtension is configured. Previously LdsServer
  hard-coded ["LDS"] regardless of whether multicast was running, so
  FindServersOnNetwork peers could not distinguish LDS from LDS-ME
  instances. Refactored the capability computation into a public static
  ComputeServerCapabilities(existing, hasMulticast) helper:
    - empty input + no multicast -> ["LDS"]
    - empty input + multicast -> ["LDS", "LDS-ME"]
    - custom seed list with no "LDS" -> "LDS" is appended
    - existing "LDS-ME" is not duplicated
  OnServerStarting now wires up the multicast layer first so the
  capability set reflects the configured topology.

* Annex C / rcp+ reverse-connect preservation: MulticastDiscovery now
  carries the OPC 10000-12 6.5.5 rcp+ DiscoveryUrl prefix end-to-end
  through mDNS announce / discover.
    - TryBuildProfile strips the rcp+ prefix before parsing the
      underlying transport URL (Uri parsing on "rcp+opc.tcp://..." is
      brittle across runtimes) and emits a dedicated TXT-record key
      ReverseConnectTxtKey ("rc") so receivers can reconstruct it
      verbatim.
    - OnServiceInstanceDiscovered honours the "rc" key when present
      and rebuilds the DiscoveryUrl with the rcp+ scheme.
  Exposes MulticastDiscovery.OpcUaServiceType,
  MulticastDiscovery.ReverseConnectScheme and
  MulticastDiscovery.ReverseConnectTxtKey as public constants so
  client code and tests can reference the same values used during
  registration and discovery.

* RegisterServer2 fallback (4.2.2): verified that LdsServer overrides
  both RegisterServerAsync and RegisterServer2Async, so the LDS
  accepts both legacy RegisterServer registrations and the modern
  RegisterServer2 + MdnsDiscoveryConfiguration path; no code change
  required.

* Tests: new LdsServerStaticTests fixture exercising the capability
  computation (six matrix cases) and locking in the Annex C / rcp+
  constants (one regression test). 195 unit + 420 integration GDS
  tests pass on net10.0; build clean on net48 and net10.0.

* Added Opc.Ua.Lds.Server as a ProjectReference of Opc.Ua.Gds.Tests so
  the new tests can use the public static surface of LdsServer and
  MulticastDiscovery.

* GDS: Push management hardening (Part 12 7.10)

* d-push-client-multitype: Make ServerPushConfigurationClient.ApplicationCertificateType
  settable instead of hard-coded to RsaSha256ApplicationCertificateType.
  Callers managing ECC, Brainpool or HTTPS certificates can now override
  the default before invoking certificate-management methods (or pass an
  explicit certificate type to the methods themselves). Removed the
  TODO comment that flagged this limitation.

* d-trustlist-ext: Set Writable / UserWritable to true for the
  DefaultApplicationGroup / DefaultHttpsGroup / DefaultUserTokenGroup
  TrustList instances in the GDS. Per OPC 10000-12 7.8.2.1 a TrustList
  that supports CloseAndUpdate / AddCertificate / RemoveCertificate is
  writeable; role-based access controls on the individual methods still
  enforce who may actually mutate the trust list. The previous values
  (false) were misleading to clients browsing the trust-list nodes.

* d-server-cred-audit: Redact the private key from
  CertificateUpdateRequestedAuditEventType InputArguments per OPC 10000-12
  7.10.3. ConfigurationNodeManager.UpdateCertificateAsync was passing
  the raw private-key ByteString into the audit payload, leaking the
  secret to any downstream audit sink. The audit payload now uses
  Opc.Ua.Server.AuditEvents.RedactedPrivateKey (an empty ByteString)
  as the placeholder; the certificate, issuer chain and key format
  remain in the payload so administrators can correlate the request.

  Also extended Server.ReportCertificateUpdateRequestedAuditEvent with
  an optional Exception parameter so callers can raise the event with
  Status=false on failure (parity with the GDS-side helpers).

* Build clean on net48 and net10.0. 195 unit + 420 integration GDS
  tests pass on net10.0.

Remaining Track D items (d-create-selfsigned, d-confirm-update,
d-managed-apps, d-alarms, d-push-tests) are larger features that
depend on Track H model additions and are not in scope for this
commit; their todos remain pending.

* GDS: Lock in ServerPushConfigurationClient.ApplicationCertificateType behaviour

Adds ServerPushConfigurationClientStaticTests covering the
Track D quick-win (d-push-client-multitype): the property defaults
to RsaSha256ApplicationCertificateType for backwards compatibility
and is now caller-overridable (e.g. for ECC management).

* GDS: Implement CreateSelfSignedCertificate (Part 12 7.10.6)

* Server-side: ConfigurationNodeManager wires the generated
  CreateSelfSignedCertificateMethodState.OnCallAsync handler when the
  optional method node exists on the ServerConfiguration instance.
  The handler:
  - Requires SecurityAdmin access (HasApplicationSecureAdminAccess)
  - Validates CertificateGroupId / CertificateTypeId via the existing
    VerifyGroupAndTypeId helper
  - Creates a self-signed certificate using the CertificateFactory
    pipeline (RSA with caller-specified key size, or ECC from the
    curve mapped by the certificateTypeId)
  - Merges DnsNames and IpAddresses into the SAN extension
  - Stores the resulting raw data on the existing
    CertificateIdentifier so the resolver can materialise it
  - Returns the DER-encoded public certificate to the caller
  The method is Optional per the spec; servers that load a nodeset
  without it simply skip the wire-up (null guard).

* Client-side: IServerPushConfigurationClient extended with
  CreateSelfSignedCertificateAsync; ServerPushConfigurationClient
  delegates to the generated ServerConfigurationTypeClient proxy
  with the standard elevate/revert permission pattern.

* net48 compat: uses new ValueTask<T>(result) instead of
  ValueTask.FromResult which is unavailable on .NET Framework 4.8.
2026-05-23 08:07:28 +02:00
Marc Schier 3e5f129c6f Enable nullable in Stack, Libraries (9), and Applications projects (#3732) 2026-05-17 12:55:24 +02:00
romanett a255facf68 [Server] Implement DiagnosticsNodeManager using AsyncCustomNodemanager (#3595)
* Refactor MonitoredItem.ValueChanged & cert validation

Refactored MonitoredItem.ValueChanged for correctness, performance, and comprehensive test coverage, including improved deadband and array handling. Rewrote equality logic for Variants and arrays. Made certificate validation thread-safe using ConcurrentDictionary and proper semaphore usage. Added extensive unit tests and benchmarks for value change detection.

* Implement DiagnosticsNodeManager using AsyncCustomNodeManager

* Implement CoreNodeManager using AsyncCustomNodeManager
2026-03-09 21:13:36 +01:00
588b7e9237 [Server] Implement event-driven RolePermissionsValidation caching in MonitoredNode2 (#3583)
* Initial plan

* Implement event-driven RolePermissionsValidation with permission caching in MonitoredNode2

Co-authored-by: romanett <7413710+romanett@users.noreply.github.com>

* Add DefaultPermissionsChanged event to IConfigurationNodeManager to handle namespace-level default permission changes

Co-authored-by: romanett <7413710+romanett@users.noreply.github.com>

* Add test & make validation functional

* cover case when DefaultRolePermissions are aded after the node exists

* cleanup

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: romanett <7413710+romanett@users.noreply.github.com>
Co-authored-by: Roman Ettlinger <romanett98@gmail.com>
2026-02-28 07:44:19 +01:00
romanett 0a4c87cef8 [Server] Implement smaller Performance improvements found when profiling (#3584)
* Add namespace index lookup for NamespaceMetadataState

Added GetNamespaceMetadataState(ushort) to ConfigurationNodeManager and interface for efficient lookup by namespace index. Switched to ConcurrentDictionary for thread safety. Updated AsyncCustomNodeManager to use index-based lookup. Cleared index cache in Clear method.

* Refactor Variant and ServiceResult handling for speed

- Use WrappedValue assignment for DataValue in MonitoredNode and NodeState to align with updated Variant usage.
- Return ServiceResult objects instead of raw StatusCodes for bad attribute IDs.
- Update ServiceResult exception handling: include stack trace only in DEBUG, message only in release.
- Optimize StatusCode to ServiceResult conversion by returning static Good/Bad instances.

* Add Read Limited load test

* Optimize MonitoredItem.IsReadToPublish

* Add Subscription unit tests and improve triggering logic

Added SubscriptionTests to verify PublishTimerExpired behavior, including timer expiry, monitored item readiness, keep-alive, and triggering scenarios. Updated Subscription.cs only process triggering if triggered items are configured, improving correctness and efficiency.

* CR feedback
2026-02-28 07:43:51 +01:00
romanett 8c008a172f [Server] Add INodeManager3 with role permission validation. Improve MonitoredNode Datastructures & Locks (#3553)
* Use Interfaces Introduced with 1.5.378

Refactor to use interfaces for node managers and server

Replaced concrete types with interfaces (e.g., IMasterNodeManager, ICoreNodeManager, IDiagnosticsNodeManager, IConfigurationNodeManager, IServerBase) across method signatures, properties, and internal structures to improve abstraction and flexibility.

* Refactor CoreNodeManager to be Based on CustomNodeManager2

* Add Test

* Add INodeManager3 with role permission validation

Introduce INodeManager3 interface extending INodeManager2, adding synchronous role permission validation methods. Update all node manager interfaces and classes to use INodeManager3. Extend IAsyncNodeManager with async permission validation methods and update adapters for sync/async bridging. Deprecate FindNodeInAddressSpace in favor of new async FindNodeInAddressSpaceAsync. Refactor MonitoredNode2 to use thread-safe collections and locking for monitored items. Enhance event reporting with security and permission checks. Update constructors and documentation to reflect new interfaces and best practices.

* fix comment

* fix
2026-02-17 17:38:36 +01:00
romanett 8b11d2bfa8 [Server] Add several interfaces to improve testability & prepare for Dependency Injection (#3448)
* add server interfaces

* Add IConfigurationNodeManager, IDiagnosticsNodeManager, IMasterNodeManager, IStandardServer, IApplicationInstance interfaces to Server. Add IMainNodeManagerFactory to create Diagnostics & MainNodeManager

* Use IApplicationInstance interface where appropriate

* fix review feedback
2026-01-12 08:06:12 +01:00