397d3c5c4f
Rename across every client surface using each language's idiomatic convention:
* .NET clients/dotnet/MxGateway.Client[.Cli|.Tests]/
-> clients/dotnet/ZB.MOM.WW.MxGateway.Client[.Cli|.Tests]/
namespaces -> ZB.MOM.WW.MxGateway.Client[.Cli|.Tests]
contracts ProjectReference repointed to ZB.MOM.WW.MxGateway.Contracts
sln migrated to slnx (dotnet sln migrate)
* Python src/mxgateway -> src/zb_mom_ww_mxgateway
src/mxgateway_cli -> src/zb_mom_ww_mxgateway_cli
distribution: mxaccess-gateway-client -> zb-mom-ww-mxaccess-gateway-client
* Rust crate: mxgateway-client -> zb-mom-ww-mxgateway-client
build.rs proto path repointed
* Java subprojects: mxgateway-{client,cli} -> zb-mom-ww-mxgateway-{client,cli}
packages com.dohertylan.mxgateway -> com.zb.mom.ww.mxgateway
group com.dohertylan.mxgateway -> com.zb.mom.ww.mxgateway
rootProject mxaccessgw-java -> zb-mom-ww-mxaccessgw-java
* Go generate-proto.ps1 proto path repointed; module path and
package mxgateway kept (Go convention).
* proto-inputs.json: generatedOutputs.python updated to new package path.
* scripts/run-client-e2e-tests.ps1: Java CLI install path + gradle task
updated to zb-mom-ww-mxgateway-cli.
CLI binary names (mxgw, mxgw-py, mxgw-go, mxgateway-cli) and wire-level
identifiers (MXGATEWAY_* env vars, the mxgw_<id>_<secret> API key
prefix, protobuf package names like mxaccess_gateway.v1, all MXAccess
references) intentionally NOT renamed.
Fix pre-existing alarms-over-gateway breaks unblocked by the rename:
* mxaccess_gateway.proto: add missing public message QueryActiveAlarmsRequest
{session_id, client_correlation_id, alarm_filter_prefix} and missing
rpc QueryActiveAlarms(QueryActiveAlarmsRequest) returns
(stream ActiveAlarmSnapshot). All four typed clients referenced
these but they were absent from the proto.
* MxAccessGatewayService.QueryActiveAlarms: implement the new RPC on
the server, streaming from IGatewayAlarmService.CurrentAlarms with
optional alarm_filter_prefix filter.
* clients/dotnet/.../DiscoverHierarchyOptions.cs: add the hand-written
.NET POCO that wraps DiscoverHierarchyRequest (referenced by
GalaxyRepositoryClient.DiscoverHierarchyAsync but never authored).
* Drop retired session_id field references from
AcknowledgeAlarmRequest/AcknowledgeAlarmReply test fixtures across
.NET, Rust, Go, and Python clients.
* Rust integration test: add the missing stream_alarms impl on the
fake MxAccessGateway server (the trait gained the method, fake
didn't).
* Rust CLI test: bump expected gatewayProtocolVersion 2 -> 3.
Regenerated artifacts updated in this commit:
* src/ZB.MOM.WW.MxGateway.Contracts/Generated/{MxaccessGateway,MxaccessGatewayGrpc}.cs
* clients/python/src/zb_mom_ww_mxgateway/generated/*_pb2{,_grpc}.py
* clients/go/internal/generated/*.pb.go
(C# regenerated by Grpc.Tools on contracts build; Python and Go via
their generate-proto.ps1 scripts; Rust regenerates from .proto via
tonic-build at compile time so no checked-in artefact.)
Verification: 472 server tests, 275 worker tests (9 dev-rig skipped),
18 integration tests (live MxAccess + LDAP + Galaxy), 57 .NET client
tests, 32 Rust workspace tests, 39 Python tests, all Go packages, and
gradle build for Java all pass.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
69 lines
2.6 KiB
C#
69 lines
2.6 KiB
C#
using Grpc.Core;
|
|
using Microsoft.Extensions.Logging;
|
|
using ZB.MOM.WW.MxGateway.Contracts.Proto;
|
|
using Polly;
|
|
using Polly.Retry;
|
|
|
|
namespace ZB.MOM.WW.MxGateway.Client;
|
|
|
|
/// <summary>Factory and helpers for exponential-backoff retry policies on transient gRPC failures.</summary>
|
|
internal static class MxGatewayClientRetryPolicy
|
|
{
|
|
/// <summary>Creates a Polly ResiliencePipeline that retries transient gRPC failures with exponential backoff.</summary>
|
|
/// <param name="options">Retry configuration (max attempts, delay bounds, jitter).</param>
|
|
/// <param name="logger">Optional logger for retry diagnostics.</param>
|
|
public static ResiliencePipeline Create(
|
|
MxGatewayClientRetryOptions options,
|
|
ILogger? logger)
|
|
{
|
|
ArgumentNullException.ThrowIfNull(options);
|
|
options.Validate();
|
|
|
|
return new ResiliencePipelineBuilder()
|
|
.AddRetry(new RetryStrategyOptions
|
|
{
|
|
MaxRetryAttempts = Math.Max(0, options.MaxAttempts - 1),
|
|
BackoffType = DelayBackoffType.Exponential,
|
|
UseJitter = options.UseJitter,
|
|
Delay = options.Delay,
|
|
MaxDelay = options.MaxDelay,
|
|
ShouldHandle = new PredicateBuilder().Handle<Exception>(IsTransientGrpcFailure),
|
|
OnRetry = args =>
|
|
{
|
|
logger?.LogDebug(
|
|
args.Outcome.Exception,
|
|
"Retrying MXAccess Gateway client call after transient gRPC failure. Attempt {Attempt}.",
|
|
args.AttemptNumber + 1);
|
|
return default;
|
|
},
|
|
})
|
|
.Build();
|
|
}
|
|
|
|
/// <summary>Returns whether a command kind is eligible for automatic retry on transient failures.</summary>
|
|
/// <param name="kind">The command kind to check.</param>
|
|
public static bool IsRetryableCommand(MxCommandKind kind)
|
|
{
|
|
return kind is MxCommandKind.Ping
|
|
or MxCommandKind.GetSessionState
|
|
or MxCommandKind.GetWorkerInfo;
|
|
}
|
|
|
|
private static bool IsTransientGrpcFailure(Exception exception)
|
|
{
|
|
return exception switch
|
|
{
|
|
RpcException rpcException => IsTransientStatus(rpcException.StatusCode),
|
|
MxGatewayException { InnerException: RpcException rpcException } => IsTransientStatus(rpcException.StatusCode),
|
|
_ => false,
|
|
};
|
|
}
|
|
|
|
private static bool IsTransientStatus(StatusCode statusCode)
|
|
{
|
|
return statusCode is StatusCode.Unavailable
|
|
or StatusCode.DeadlineExceeded
|
|
or StatusCode.ResourceExhausted;
|
|
}
|
|
}
|