GithubStar/UA-.NETStandard
1
0
Fork 0
mirror of https://github.com/OPCFoundation/UA-.NETStandard.git synced 2026-10-16 21:02:51 +08:00
Code Issues Packages Projects Releases Wiki Activity
Files
master
UA-.NETStandard/Directory.Packages.props
T
romanett 28f1401324 Server: cut request-path lock contention and allocation; add load harness (#4707) 
Profiling the reference server under load showed most request-path CPU
spent spinning on contended locks, plus several avoidable allocations
per request or per node. This PR removes them, switches the reference
server to Server GC, and adds the load harness used to measure it.

## Results

Paired A/B against `master` (3 alternating rounds, medians, 20 sessions,
workstation GC on both sides, 24-core machine):

| Scenario | Throughput | p50 | p90 | Server CPU / op | Server alloc /
op | Lock contention |
|---|---|---|---|---|---|---|
| Read, 1 value | **+45%** (94k → 137k req/s) | −20% | −46% | −37% |
−39% | −94% |
| Read, 100 values | **+40%** | −26% | −17% | −24% | −25% | +130% ¹ |
| Read, 100 values, SignAndEncrypt | **+48%** | −36% | −20% | −30% |
−26% | +106% ¹ |
| Browse (Server object) | **+33%** | −31% | −12% | −13% | −14% | −40% |
| Write, 100 values | +13% | −12% | −4% | −10% | −19% | −23% |
| Data-change subscriptions | ±0 ² | ≈ | ≈ | −14% | −17% | −29% |

¹ The faster server spends more time in the remaining per-node locks.
² Throughput is bounded by the writer rate in this scenario.

Separately, Server GC instead of workstation GC doubles Read-100
throughput (21k → 44k req/s) and cuts p90 from 2.1 ms to 0.6 ms, because
workstation GC paused the server for about half the wall time under
load. The working set grows from ~210 MB to ~430 MB.

## Changes

- **`ObjectPool<T>`**: lock-free slot array. `ConcurrentBag.Count` took
every per-thread lock on each `Return` of the pooled response value-task
source, once per request.
- **`FairRequestQueue`**:
- drops redundant gate entries: the fail-fast pre-checks and the stopped
check read volatile fields and are re-checked under the gate; the owner
count is folded into selection;
- coalesces wake pulses with a pending flag, so the bounded signal
channel's lock is only taken when no token is outstanding. Wake-count
semantics in the tests are unchanged.
- **`DefaultServerResourceIsolationProvider`**:
  - compares peer addresses without `GetAddressBytes()` allocations;
- builds the unsecured verification context only when a classifier is
configured.
- **TCP listener / channel**: caches the certificate DER copies per
channel, keyed on the certificate instance. `X509Certificate2.RawData`
returns a fresh copy on every access, and the listener read it for every
request.
- **`EndpointBase`**: resolves the request `ActivitySource` once instead
of `Assembly.GetCallingAssembly()` per request.
- **Node managers**:
- one `ServerSystemContext` copy per request (cached on the
`OperationContext`) for per-node callbacks: permission metadata, node
metadata, view membership, method lookup;
  - removes a discarded copy in `GetReferenceDescriptionAsync`;
- per-request operation caches are plain `Dictionary` instead of a
`ConcurrentDictionary` with 24 lock objects;
  - the validation cache is built in one pass;
  - fixes zero-capacity error lists.
- **`ResourceManager`**: translation lookups read an immutable snapshot
that is rebuilt after writes. Browse serialised on the global lock for
every display name.
- **`MonitoredItem`**: shares the default `DataChangeFilter` instead of
allocating one per sampled value.
- **Reference server**: `ServerGarbageCollection=true`; new "Garbage
collector mode" section in `docs/ServerScalability.md`.
- **`perf/`**: `ServerLoadHarness` (out-of-process server plus load
driver, server-side CPU/GC/alloc/contention deltas, optional
`dotnet-trace` capture), `TraceAnalyzer` (TraceEvent-based
CPU/allocation/contention aggregation), and paired A/B scripts. Not part
of `UA.slnx`.

### Behaviour notes for custom node managers
- The `IDictionary<NodeId, NodeState>` operation cache handed to
overrides is now a `Dictionary`. It was a `ConcurrentDictionary`, but
only one request flow uses it.
- Within one request, per-node callbacks of a node manager share one
`ServerSystemContext` copy.

### Later commits on this PR
- **Permission checks without `NodeMetadata`** (`9068684c3`): the
role-permission and access-restriction validators run on an internal
`PermissionMetadata` struct; `AsyncCustomNodeManager` fills it on the
stack from the same attributes and namespace defaults. Browse's
reference filter reads 3 attributes per target instead of 13. Read 100
values: 120.6 → 95.7 KB/request, ~21k → 26k req/s; Browse 67.6 → 51.8
KB, 26.7k → 29.8k req/s.
- **Read-only channel evidence** (`936b28d76`, `dc638ab9f`):
`SecureChannelContext.ClientChannelCertificate`,
`ServerChannelCertificate` and `ChannelThumbprint` are `ByteString`, so
the cached per-channel certificate buffer is shared without a copy and
cannot be altered by handlers or classifiers. **Public API change**, see
the new section in `docs/MigrationGuide.md`. The isolation provider's
revalidation snapshot still copies.
- **Session lookup** (`936b28d76`):
`SessionManager.TryGetSessionContext` takes the candidate snapshot
without the bindings lock; the validating check under the lock is
unchanged (2 instead of 4 global lock acquisitions per request).
- Unit tests for `ObjectPool`, the per-request system context cache, the
translation snapshot, `SecureChannelContext` null semantics and a
stopped `FairRequestQueue`; `perf/` follows the repository conventions.

## Testing
- `Opc.Ua.Server.Tests` (net10.0): 8386 passed before the `ByteString`
change. After it, 13 secure-session and isolation tests failed; both
causes are fixed in `dc638ab9f` and those tests (25 with their variants)
pass. A full rerun of the Server, Sessions, Core, Redundancy, LDS and
GDS suites on the final commit is in progress.
- Full solution builds for net10.0 and net48; `dotnet format whitespace`
clean.

### o6-automation opcua-benchmarks (sampling, deep pipelines)
Replicated both tests of the [o6-automation
benchmark](https://github.com/o6-automation/opcua-benchmarks) in the
harness (`--scenario counters`, and `read` with 10 clients × N
outstanding), paired against `origin/master`:

| Test | master | this PR |
|---|---|---|
| Subscription sampling, 100 Hz, every item ≥ 99 % of samples, 1 item |
FAIL (75–94 %) | PASS (100 %) |
| same, 1024 items | FAIL (64–73 %) | PASS (99.9 %) |
| same, capacity | — | between 4096 and 6144 items (~410–520k samples/s)
|
| Read, 10 clients × 100 outstanding, None | stalls (0–1.7k req/s) |
121–131k req/s |
| Read, 10 clients × 100 outstanding, SignAndEncrypt | stalls (0 req/s)
| 89–91k req/s |
| Read, 10 clients × 1000 outstanding | channels faulted
(`BadTcpNotEnoughResources`) | 0 errors, ~45–54k req/s |

- **Fixed-rate sampling** (`30b769d13`): the sampling loop waited one
interval after each sample, adding timer overshoot and sampling time to
every cycle; samples now have absolute deadlines.
- **Async response send** (`cdac5c5ca`): `SendResponse` blocked
thread-pool workers on the channel gate; with many requests in flight
the receive loop that was handed the gate waited for a pool thread and
the channel stalled. The TCP, Kestrel TCP and WebSocket listeners now
use `SendResponseAsync`.
- **Backpressure instead of channel fault** (`191cd2ba7`): server
channels stop reading while more than 100 responses are queued and
resume at 50, instead of faulting the channel. Client channels keep
reading. At 10,000 outstanding requests throughput is ~50k req/s with or
without the limit engaged, so the remaining gap to the 100-per-client
case is not the backpressure.

## Not changed
- `SentMessageQueue` deep-clones every `NotificationMessage`. Retained
messages return their pooled payloads on acknowledgement while the
response copy may still be queued for send, and a client can acknowledge
a sequence number before receiving it; the clone keeps a recycled
payload from reaching another session's response. Removing it needs a
send-completion hook from the transport.
- Generated responses allocate a default `ResponseHeader` that is
replaced (~2-3 % of per-request allocation). Fixing it means making
structure fields of every generated type lazily initialized, a
generator-wide change to default values, equality and encoding.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-09 13:00:49 +02:00

208 lines
14 KiB
XML
Raw Permalink Blame History

<Project>
<!-- Roslyn API versions the analyzers and source generators are built against.
The central pins below track the newest of them so that the primary
analyzer variants and the projects that test them resolve identically. -->
<Import Project="roslyn.props" />
<PropertyGroup>
<ManagePackageVersionsCentrally>true</ManagePackageVersionsCentrally>
<!-- Pin transitive package versions so PackageVersion entries below force the
resolved version everywhere they appear in the dependency graph, even when
no project declares a direct PackageReference. Used to remediate transitive
NU1903 vulnerabilities (System.Net.Http 4.3.0, System.Text.RegularExpressions
4.3.0) pulled in by Microsoft.AspNetCore.* 2.3.x on net8.0/net9.0. -->
<CentralPackageTransitivePinningEnabled>true</CentralPackageTransitivePinningEnabled>
</PropertyGroup>
<ItemGroup>
<PackageVersion Include="BenchmarkDotNet" Version="0.15.8" />
<PackageVersion Include="Microsoft.Diagnostics.Tracing.TraceEvent" Version="3.1.21" />
<PackageVersion Include="BitFaster.Caching" Version="2.6.1" />
<PackageVersion Include="BouncyCastle.Cryptography" Version="2.7.0" />
<PackageVersion Include="Confluent.Kafka" Version="2.15.1" />
<PackageVersion Include="coverlet.collector" Version="10.0.1" />
<PackageVersion Include="Crdt" Version="1.1.2" />
<PackageVersion Include="Crdt.Transport" Version="1.1.2" />
<PackageVersion Include="Dekaf" Version="1.21.1" />
<PackageVersion Include="RaftCs" Version="1.1.2" />
<PackageVersion Include="RaftCs.Transport" Version="1.1.2" />
<PackageVersion Include="RaftCs.Transport.NanoMsg" Version="1.1.2" />
<PackageVersion Include="RaftCs.Storage.File" Version="1.1.2" />
<PackageVersion Include="EmbedIO" Version="3.5.2" />
<PackageVersion Include="JsonSchema.Net" Version="9.4.0" />
<PackageVersion Include="Makaretu.Dns.Multicast" Version="0.27.0" />
<PackageVersion Include="Microsoft.AspNetCore.Connections.Abstractions" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Hosting" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Http" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Server.Kestrel" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Server.Kestrel.Core" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Server.Kestrel.Https" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.Server.Kestrel.Transport.Sockets" Version="2.3.13" />
<PackageVersion Include="Microsoft.AspNetCore.WebSockets" Version="2.3.13" />
<PackageVersion Include="Microsoft.Bcl.HashCode" Version="6.0.0" />
<PackageVersion Include="Microsoft.Bcl.AsyncInterfaces" Version="10.0.12" />
<PackageVersion Include="Microsoft.Bcl.TimeProvider" Version="10.0.12" />
<PackageVersion Include="Microsoft.CodeAnalysis.Analyzers" Version="5.9.0" />
<PackageVersion Include="Microsoft.CodeAnalysis.Common" Version="$(RoslynApiVersionVS2026)" />
<PackageVersion Include="Microsoft.CodeAnalysis.CSharp" Version="$(RoslynApiVersionVS2026)" />
<PackageVersion Include="Microsoft.CodeAnalysis.CSharp.Workspaces" Version="$(RoslynApiVersionVS2026)" />
<PackageVersion Include="Microsoft.CodeAnalysis.Workspaces.Common" Version="$(RoslynApiVersionVS2026)" />
<PackageVersion Include="Microsoft.Extensions.Configuration" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Configuration.Abstractions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Configuration.EnvironmentVariables" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Logging" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Logging.Abstractions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Logging.Console" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.DependencyInjection.Abstractions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.DependencyInjection" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Diagnostics" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Http" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Http.Resilience" Version="10.10.0" />
<PackageVersion Include="Microsoft.Extensions.Hosting" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Hosting.Abstractions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Caching.Abstractions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Options" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.Options.ConfigurationExtensions" Version="10.0.12" />
<PackageVersion Include="Microsoft.Extensions.TimeProvider.Testing" Version="10.10.0" />
<PackageVersion Include="ModelContextProtocol" Version="2.2.0" />
<PackageVersion Include="ModelContextProtocol.AspNetCore" Version="2.2.0" />
<PackageVersion Include="ModelContextProtocol.Core" Version="2.2.0" />
<PackageVersion Include="Microsoft.NET.Test.Sdk" Version="18.10.1" />
<PackageVersion Include="Microsoft.SourceLink.AzureRepos.Git" Version="10.0.401" />
<!-- AI Model Management. Microsoft.Extensions.AI is the inference
abstraction: IChatClient is the contract that reaches a hosted service
and an on-device runtime alike, which is the property the specification
asserts about where inference runs. It pins System.Text.Json 10.0.12,
the version this repository already centralises on, so it costs nothing
to take. No cloud-vendor SDK is referenced anywhere: workload identity
is read from the token the platform projects, which is the mechanism
every platform implements underneath its own SDK. -->
<PackageVersion Include="Microsoft.Extensions.AI" Version="10.10.0" />
<PackageVersion Include="Microsoft.SourceLink.GitHub" Version="10.0.401" />
<PackageVersion Include="System.CommandLine" Version="2.0.12" />
<PackageVersion Include="Microsoft.VisualStudio.DiagnosticsHub.BenchmarkDotNetDiagnosers" Version="18.7.37220.1" />
<PackageVersion Include="Moq" Version="4.20.72" />
<PackageVersion Include="TUnit" Version="1.69.0" />
<PackageVersion Include="Nerdbank.GitVersioning" Version="3.10.94" />
<PackageVersion Include="Newtonsoft.Json" Version="13.0.4" />
<!--
PacketDotNet and SharpPcap are native-code dependencies of
Opc.Ua.Core.Diagnostics, Opc.Ua.PubSub.Diagnostics and (for the
opt-in WithPcap() backend, net8.0+ only) Opc.Ua.PubSub.Eth. They
dynamically load libpcap on Linux/macOS and Npcap on Windows. Per
SDL native-code policy:
- Versions are pinned (no floating range)
- SBOM tracking is required for every release
- Subscribe to upstream CVE feeds:
* https://github.com/dotpcap/sharppcap/security/advisories
* https://github.com/dotpcap/packetnet/security/advisories
- Review monthly; bump within 30 days for HIGH/CRITICAL CVEs
-->
<PackageVersion Include="PacketDotNet" Version="1.4.8" />
<PackageVersion Include="Pkcs11Interop" Version="5.3.0" />
<PackageVersion Include="SharpPcap" Version="6.3.1" />
<PackageVersion Include="NUnit" Version="4.6.1" />
<PackageVersion Include="NUnit.Analyzers" Version="4.15.0" />
<PackageVersion Include="NUnit.Console" Version="3.22.0" />
<PackageVersion Include="NUnit3TestAdapter" Version="6.3.0" />
<PackageVersion Include="ObjectLayoutInspector" Version="0.2.0" />
<PackageVersion Include="PolySharp" Version="1.16.0" />
<PackageVersion Include="Roslynator.Analyzers" Version="5.0.0" />
<PackageVersion Include="Roslynator.Formatting.Analyzers" Version="5.0.0" />
<PackageVersion Include="SharpFuzz" Version="2.3.0" />
<!-- Transitive vulnerability pin (consumed via Testcontainers.*):
- SSH.NET 2026.0.0 fixes GHSA-q939-rpr3-3284 -->
<PackageVersion Include="SSH.NET" Version="2026.0.0" />
<PackageVersion Include="Serilog" Version="4.4.0" />
<PackageVersion Include="Testcontainers" Version="4.15.0" />
<PackageVersion Include="Testcontainers.Kafka" Version="4.15.0" />
<PackageVersion Include="Serilog.Expressions" Version="5.0.0" />
<PackageVersion Include="Serilog.Extensions.Logging" Version="10.0.0" />
<PackageVersion Include="Serilog.Sinks.Console" Version="6.1.1" />
<PackageVersion Include="Serilog.Sinks.Debug" Version="3.0.0" />
<PackageVersion Include="Serilog.Sinks.File" Version="7.0.0" />
<PackageVersion Include="System.Buffers" Version="4.6.1" />
<PackageVersion Include="System.Composition.AttributedModel" Version="10.0.12" />
<PackageVersion Include="System.Runtime.CompilerServices.Unsafe" Version="6.1.2" />
<PackageVersion Include="System.Diagnostics.DiagnosticSource" Version="10.0.12" />
<PackageVersion Include="System.Memory" Version="4.6.3" />
<PackageVersion Include="System.Formats.Asn1" Version="10.0.12" />
<PackageVersion Include="System.IO.Compression" Version="4.3.0" />
<PackageVersion Include="System.IO.FileSystem.AccessControl" Version="5.0.0" />
<!-- Transitive vulnerability pins (consumed via Microsoft.AspNetCore.* 2.3.x on
net8.0/net9.0 - default transitives are 4.3.0 with NU1903):
- System.Net.Http 4.3.4 fixes GHSA-7jgj-8wvc-jh57
- System.Text.RegularExpressions 4.3.1 fixes GHSA-cmhx-cq75-c4mj -->
<PackageVersion Include="System.Net.Http" Version="4.3.4" />
<PackageVersion Include="System.Text.RegularExpressions" Version="4.3.1" />
<PackageVersion Include="System.Collections.Immutable" Version="$(RoslynRuntimeVersion)" />
<PackageVersion Include="System.Reflection.Metadata" Version="$(RoslynRuntimeVersion)" />
<PackageVersion Include="System.Text.Json" Version="10.0.12" />
<PackageVersion Include="System.Private.Uri" Version="4.3.2" />
<PackageVersion Include="System.Runtime.InteropServices.RuntimeInformation" Version="4.3.0" />
<PackageVersion Include="System.Security.Cryptography.Cng" Version="5.0.0" />
<PackageVersion Include="System.Threading.Channels" Version="10.0.12" />
<PackageVersion Include="System.Threading.RateLimiting" Version="10.0.12" />
<PackageVersion Include="System.Threading.Tasks.Extensions" Version="4.6.3" />
</ItemGroup>
<!-- OpenUSD rendering stack, consumed by the optional
tools/Opc.Ua.OpenUsd.Connector.Viewer assembly that backs the connector's
view option, and by src/Opc.Ua.Vision.OpenUsd which renders the
Vision server's simulated camera views offscreen via Silk (D3D12/Vulkan).
Published on nuget.org. The runtime packages ship win-x64, linux-x64 and
osx-arm64; the RID-agnostic metapackages resolve the correct native asset
per RID, so no RID-conditional reference is required. All of these move
together because the managed and RID-specific runtime packages have to
match; a partial bump fails when the natives load rather than when the
solution builds. -->
<ItemGroup>
<PackageVersion Include="OpenUsd" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Viewer" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Rendering" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Rendering.Silk" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Rendering.Silk.D3D12" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Rendering.Silk.Vulkan" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Runtime.Core" Version="0.14.0-alpha" />
<PackageVersion Include="OpenUsd.Runtime.Imaging" Version="0.14.0-alpha" />
<PackageVersion Include="Avalonia" Version="12.1.3" />
<PackageVersion Include="Avalonia.Desktop" Version="12.1.3" />
</ItemGroup>
<!-- MQTTnet 5.x dropped net48 support. Per-TFM pinning is
required because CentralPackageTransitivePinningEnabled lets the central pin
win over PubSub's VersionOverride for transitive consumers (e.g. samples
that ProjectReference Opc.Ua.PubSub). See src/Opc.Ua.PubSub/Opc.Ua.PubSub.csproj. -->
<ItemGroup Condition="'$(TargetFramework)' == 'net48'">
<PackageVersion Include="MQTTnet" Version="4.3.7.1207" />
</ItemGroup>
<ItemGroup Condition="'$(TargetFramework)' != 'net48'">
<PackageVersion Include="MQTTnet" Version="5.2.0.1603" />
<PackageVersion Include="MQTTnet.Server" Version="5.2.0.1603" />
</ItemGroup>
<!-- ASP.NET Core packages that ship one band per .NET major. Unlike the
Microsoft.Extensions.* packages above, these have no netstandard2.0 asset and
no roll-forward across majors: a net8.0 project cannot consume the 10.0.x
band. The consumers (src/Opc.Ua.Bindings.Https and
tests/Opc.Ua.Bindings.Https.WebApi.Tests) build for net8.0, net9.0 and
net10.0, so the version has to follow $(TargetFramework) rather than being a
single pin held down to the oldest supported band.
The net10.0 group is the fallback so that a TFM this file does not enumerate
- a future net11.0 - still resolves a version. -->
<ItemGroup Condition="'$(TargetFramework)' == 'net8.0'">
<PackageVersion Include="Microsoft.AspNetCore.Authentication.Certificate" Version="8.0.31" />
<PackageVersion Include="Microsoft.AspNetCore.Authentication.JwtBearer" Version="8.0.31" />
<PackageVersion Include="Microsoft.AspNetCore.Mvc.Testing" Version="8.0.31" />
<PackageVersion Include="Microsoft.AspNetCore.TestHost" Version="8.0.31" />
</ItemGroup>
<ItemGroup Condition="'$(TargetFramework)' == 'net9.0'">
<PackageVersion Include="Microsoft.AspNetCore.Authentication.Certificate" Version="9.0.20" />
<PackageVersion Include="Microsoft.AspNetCore.Authentication.JwtBearer" Version="9.0.20" />
<PackageVersion Include="Microsoft.AspNetCore.Mvc.Testing" Version="9.0.20" />
<PackageVersion Include="Microsoft.AspNetCore.TestHost" Version="9.0.20" />
</ItemGroup>
<ItemGroup Condition="'$(TargetFramework)' != 'net8.0' AND '$(TargetFramework)' != 'net9.0'">
<PackageVersion Include="Microsoft.AspNetCore.Authentication.Certificate" Version="10.0.12" />
<PackageVersion Include="Microsoft.AspNetCore.Authentication.JwtBearer" Version="10.0.12" />
<PackageVersion Include="Microsoft.AspNetCore.Mvc.Testing" Version="10.0.12" />
<PackageVersion Include="Microsoft.AspNetCore.TestHost" Version="10.0.12" />
</ItemGroup>
</Project>
Reference in New Issue View Git Blame Copy Permalink