fix(client-dotnet): size the safe-unary retry budget for every attempt (#197)

ExecuteSafeUnaryAsync capped the whole retry pipeline at one
DefaultCallTimeout while each attempt carried the same deadline, so a first
attempt ending in DeadlineExceeded spent the entire budget and the retry
never ran. The outer cap is now MaxAttempts x DefaultCallTimeout plus MaxDelay
per backoff (MxGatewayClientRetryPolicy.SafeUnaryBudget), in both
MxGatewayClient and GalaxyRepositoryClient; a budget too large for a
cancellation timer leaves only the per-attempt deadlines.

Tests: MxGatewayClientWireTests.SafeUnaryRetry_AfterDeadlineExceededFirstAttempt_RunsTheRetry
(real gRPC loopback server stalls the first Invoke; fails with Cancelled on
the old code), plus two SafeUnaryBudget unit tests.
This commit is contained in:
Joseph Doherty
2026-11-06 16:07:35 -05:00
parent ff397b27be
commit b9096f7c57
8 changed files with 159 additions and 7 deletions
@@ -517,8 +517,15 @@ public sealed class GalaxyRepositoryClient : IAsyncDisposable
Func<CancellationToken, Task<T>> call,
CancellationToken cancellationToken)
{
// Each attempt carries its own DefaultCallTimeout deadline, so the outer cap must cover
// every attempt plus the backoff between them; capping it at one DefaultCallTimeout let a
// DeadlineExceeded first attempt spend the whole budget and the retry never ran (#197).
using CancellationTokenSource timeout = CancellationTokenSource.CreateLinkedTokenSource(cancellationToken);
timeout.CancelAfter(Options.DefaultCallTimeout);
TimeSpan? budget = MxGatewayClientRetryPolicy.SafeUnaryBudget(Options);
if (budget is not null)
{
timeout.CancelAfter(budget.Value);
}
return await _safeUnaryRetryPipeline.ExecuteAsync(
async token => await call(token).ConfigureAwait(false),