Files
UA-.NETStandard/docs/TransferSubscription.md
Marc Schier b1c0cd20fd Fix classic browse-path subscription restoration (#4814)
# Description

Persisted classic subscriptions retain their server/client handles but
previously lost relative-path resolution metadata. Transfer could
succeed, yet a later `ApplyChangesAsync` attempted to resolve an
already-created item and failed. The SDK now owns this restoration
lifecycle without exposing a public resolution-state setter.

- Preserve the originally monitored target in internal snapshot state
and restore both `ResolvedNodeId` and `Status.NodeId`.
Transfer/reactivation use handles without re-browsing; fresh creation
resolves paths again.
- Version classic subscription/configuration files and read existing
unversioned 2.0 files using frozen legacy wire layouts. Legacy missing
targets remain unknown rather than being guessed.
- Exclude existing items from creation-time resolution, reserve creation
before asynchronous translation, and reject stale translation results.
Failed deletion retains identity and pending retry state.
- Support forward-only snapshot input using position tracking and
bounded, per-ExtensionObject buffering in the binary decoder, including
short reads. No whole-file buffering or reflection-based serialization
is introduced.
- Add unit/integration regressions and document the supported lifecycle
and migration implications.

**Compatibility:** Newly written files require the updated SDK. Existing
2.0 files remain readable. Assigning a different `StartNodeId` or
`RelativePath` to a created/in-flight item now immediately throws
`BadInvalidState`; unchanged assignments remain allowed. No V2
subscription-engine changes or 1.5 XML import are included.

## Checklist

_Put an `x` in the boxes that apply. You can complete these step by step
after opening the PR._

- [ ] 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.
- [ ] 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.
- [ ] I ran **all** tests locally using the **UA.slnx** solution against
at least .net **framework** and .net **10**, and all passed.
- [ ] I fixed **all** failing and flaky tests in the CI pipelines and
**all** CodeQL warnings.
- [ ] I have addressed **all** PR feedback received.
2026-10-16 20:06:51 +02:00

6.6 KiB

TransferSubscriptions

Contents

Overview

TransferSubscriptions moves existing server subscriptions to another session. It can reduce data loss during reconnect or client restart by preserving the server's monitored items and queued notifications. It does not guarantee zero loss: subscription lifetime, queue capacity, retransmission availability, and authorization still apply.

The server and client implement the Part 4 TransferSubscriptions service. The target session must have permission to transfer the subscriptions. Transfer also relies on the server's GetMonitoredItems method when the client needs to recover monitored-item handles.

For current client APIs:

  • Use ManagedSession recovery for connection recovery; do not add SessionReconnectHandler around a ManagedSession.
  • Use subscription persistence and transfer to save client state and restore it after restart. A transfer attempts to recover an existing server subscription; recreating one starts new queues.
  • Classic relative-path items retain their originally resolved targets in new snapshots. Adoption uses saved handles, not a fresh browse-path translation. Older snapshots with unknown target metadata still support transfer and subsequent maintenance; callers do not need to inject resolution results.
  • If subscriptions must survive a deliberate session close, configure DeleteSubscriptionsOnClose = false before closing it. They remain subject to their server-side lifetime; this does not make them durable across server restart. See durable subscriptions for that separate capability.

Porting an existing server

Servers based on StandardServer and the built-in node-manager bases already participate in monitored-item transfer. A custom manager implementing the interfaces directly must implement the transfer contract, including initial-value queuing when requested, without retaining stale source-session ownership. See node managers and the 2.0 session/subscription migration guide.

Transfer support was introduced in 1.4.368. The historical sample port is relevant only when upgrading from releases predating that support; it is not the current SDK API reference.

Known limitations and issues

  • Transfer cannot recover samples already lost to queue overflow or expired subscriptions.
  • Server restart requires server-side durability or mirroring to retain the subscription; saving client configuration alone is insufficient.
  • Recovery from an unsolicited transfer notification is an explicit compatibility policy, described below, not a substitute for legitimate cross-session transfer.

Recovering from unsolicited Good_SubscriptionTransferred

Per OPC UA Part 4 §5.14.7 the server emits a StatusChangeNotification with Good_SubscriptionTransferred on the old Session whenever the subscription is transferred away to a new Session via TransferSubscriptions. The receiving Session is expected to treat the subscription as gone and stop dispatching for it.

Some servers have been observed to deliver this notification against a subscription the client has just freshly created on its current Session — for example, Kepware after a server re-initialisation can leak the pre-restart subscription's pending status notifications onto the new session (especially noticeable because subscription identifiers are re-used starting at 1). The client then sees the per-subscription dispatch silently disabled even though, from its point of view, the subscription is alive and should keep receiving data. Tracked as issue #3540.

To opt into automatic in-place recovery, set SubscriptionRecoveryPolicy.RecreateOnUnsolicitedTransfer:

// Classic (V1) subscription
var subscription = new Subscription(telemetry, options)
{
    RecoveryPolicy = SubscriptionRecoveryPolicy.RecreateOnUnsolicitedTransfer,
    FastDataChangeCallback = OnDataChange
};
// V2 subscription options
var options = new Opc.Ua.Client.Subscriptions.SubscriptionOptions
{
    PublishingEnabled = true,
    RecoveryPolicy = SubscriptionRecoveryPolicy.RecreateOnUnsolicitedTransfer
};

When the policy is RecreateOnUnsolicitedTransfer and a Good_SubscriptionTransferred arrives while the subscription is still actively owned by this Session, the SDK will:

  1. Drop every queued acknowledgement targeting the dead subscription id — this prevents BadSubscriptionIdInvalid ack errors on servers that re-use subscription identifiers across generations.
  2. Recreate the subscription on the same Session via CreateSubscription, obtaining a fresh server-side subscription identifier.
  3. Re-issue the monitored items so the data flow resumes against the new identifier.

The default is SubscriptionRecoveryPolicy.ReportOnly, which preserves the spec-strict behaviour: the PublishStateChangedMask.Transferred flag is raised (V1) or PublishState.Transferred is dispatched (V2) so the application can react manually, and the per-subscription publish dispatch is stopped (V1) or left as a no-op (V2).

Caveats

  • Auto-recovery is not lossless failover. The server-side retransmission queue and any triggering relationships tied to the invalidated subscription identifier are lost; only the subscription's wire-level options and the configured monitored items are re-applied.
  • The recovery path runs against the same Session. If the Session itself is reconnecting, the V1 implementation defers to the reconnect pipeline rather than racing it; the V2 implementation lets the in-place recreate run when the subscription is Created.
  • The "unsolicited" classification is conservative — concurrent recovery dispatches collapse into one — but it does not attempt to detect every legitimate cross-session/cross-client failover. If your design relies on another Session/Client legitimately pulling subscriptions away via TransferSubscriptions, keep the default ReportOnly policy and handle the PublishStatusChanged event explicitly.