fix: exclude shutting down silos from placement#10219
Open
ReubenBond wants to merge 13 commits into
Open
Conversation
ReubenBond
force-pushed
the
reubenbond/active-silo-placement
branch
from
b20e126 to
f7a27ef
Compare
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Canonicalize cluster manifests and nested manifest properties so repeated metadata is shared across silo entries and AllGrainManifests contains unique manifests. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
ReubenBond
force-pushed
the
reubenbond/active-silo-placement
branch
from
da9e72d to
cdef309
Compare
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Prune cluster manifests when membership versions advance so GrainVersionManifest and version selector caches only expose active silos. Simplify placement compatibility to trust those active-only caches while retaining the active-silo guard after extensible filters. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Prune cluster manifests when membership versions advance so GrainVersionManifest and version selector caches only expose active silos. Simplify placement compatibility to trust those active-only caches while retaining the active-silo guard after extensible filters. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
ReubenBond
force-pushed
the
reubenbond/active-silo-placement
branch
from
cdef309 to
9bb75b5
Compare
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates Orleans placement and cluster-manifest handling so that placement candidates/manifests better reflect current cluster availability (notably around shutdown/termination), while also tightening APIs and reducing manifest duplication.
Changes:
- Update
IPlacementContext.GetCompatibleSilosto returnImmutableArray<SiloAddress>and adjust placement directors/tests to match. - Update
PlacementServiceto useISiloStatusOracle.GetActiveSilos()for active-silo filtering and filter versioned compatible-silo results by active membership. - Improve cluster manifest behavior (sync pruning of non-active silos, manifest publishing behavior, manifest deduplication) and add new targeted tests.
Show a summary per file
| File | Description |
|---|---|
| test/Orleans.Core.Tests/Runtime/PlacementServiceTests.cs | Adds coverage for excluding local silo during shutdown; updates test scaffolding for GetActiveSilos()/ImmutableArray. |
| test/Orleans.Core.Tests/Runtime/LocalPlacementDirectorTests.cs | Updates mocks to return ImmutableArray from GetCompatibleSilos. |
| test/Orleans.Core.Tests/Runtime/ActivationCountPlacementDirectorTests.cs | Updates mocks to return ImmutableArray from GetCompatibleSilos. |
| test/Orleans.Core.Tests/Manifest/ClusterManifestProviderTests.cs | New tests validating manifest pruning/publishing behavior and placement’s active-only results. |
| test/Orleans.Core.Tests/Manifest/ClusterManifestDeduplicationTests.cs | New tests validating structural equality/deduplication of manifest objects. |
| test/Grains/TestGrains/VersionAwarePlacementDirector.cs | Adjusts to treat compatible silos as IReadOnlyList for ImmutableArray compatibility. |
| src/Orleans.Runtime/Versions/CachedVersionSelectorManager.cs | Switches cached suitable-silos list to ImmutableArray. |
| src/Orleans.Runtime/Placement/SiloRoleBasedPlacementDirector.cs | Uses ImmutableArray for compatible-silo list and IsEmpty checks. |
| src/Orleans.Runtime/Placement/PlacementService.cs | Changes compatible-silo API to ImmutableArray, uses GetActiveSilos(), and filters versioned results by actives. |
| src/Orleans.Runtime/Placement/ActivationCountPlacementDirector.cs | Updates selector helper signature to accept ImmutableArray. |
| src/Orleans.Runtime/Manifest/ClusterManifestProvider.cs | Refactors manifest synchronization/publishing and membership-update processing; now depends on IClusterMembershipService. |
| src/Orleans.Core/Placement/IPlacementDirector.cs | Updates placement hint helper signature to accept ImmutableArray. |
| src/Orleans.Core/Placement/IPlacementContext.cs | Updates placement context API to return ImmutableArray. |
| src/Orleans.Core/Manifest/GrainBindings.cs | Small refactor to centralize manifest processing loop into a local function. |
| src/Orleans.Core/Core/GrainInterfaceTypeToGrainTypeResolver.cs | Refactor to centralize per-manifest cache building logic into a helper method. |
| src/Orleans.Core.Abstractions/Manifest/GrainProperties.cs | Adds structural equality/hash support for deduplication. |
| src/Orleans.Core.Abstractions/Manifest/GrainManifest.cs | Adds structural equality/hash support for deduplication. |
| src/Orleans.Core.Abstractions/Manifest/GrainInterfaceProperties.cs | Adds structural equality/hash support for deduplication. |
| src/Orleans.Core.Abstractions/Manifest/ClusterManifest.cs | Adds AllGrainManifests ctor and implements manifest/property deduplication. |
| src/api/Orleans.Core/Orleans.Core.cs | Public API update for GetCompatibleSilos and GetPlacementHint signatures. |
| src/api/Orleans.Core.Abstractions/Orleans.Core.Abstractions.cs | Public API update for new ClusterManifest ctor and new equality members. |
Copilot's findings
- Files reviewed: 21/21 changed files
- Comments generated: 6
Comment on lines
+192
to
196
| if (member.SiloAddress.Equals(_localSiloAddress)) | ||
| { | ||
| // If the member is not yet active, it may not be ready to process requests. | ||
| // Local membership changes are applied synchronously by EnsureCurrentManifestVersion. | ||
| continue; | ||
| } |
| compatibleSilos = filteredSilos.ToArray(); | ||
| } | ||
|
|
||
| compatibleSilos = compatibleSilos.Intersect(_siloStatusOracle.GetActiveSilos()).ToImmutableArray(); |
Comment on lines
+249
to
+252
| var activeSilos = _siloStatusOracle.GetActiveSilos(); | ||
| return silos.ToDictionary( | ||
| static entry => entry.Key, | ||
| entry => entry.Value.Intersect(activeSilos).ToArray()); |
Comment on lines
+114
to
+119
| configureOptions: options => options.AssumeHomogenousSilosForTesting = true, | ||
| configureSiloStatusOracle: (siloStatusOracle, localSilo) => | ||
| { | ||
| siloStatusOracle.CurrentStatus.Returns(SiloStatus.ShuttingDown); | ||
| siloStatusOracle.GetActiveSilos().Returns(ImmutableArray.Create(remoteSilo)); | ||
| }); |
| @@ -31,7 +32,7 @@ Task<SiloAddress> OnAddActivation( | |||
| /// <param name="requestContextData">The request context data.</param> | |||
| /// <param name="compatibleSilos">The compatible silos.</param> | |||
| /// <returns>The placement hint, if present and valid, or <see landword="null"/> otherwise.</returns> | |||
Comment on lines
+124
to
+145
| var builder = ImmutableDictionary.CreateBuilder<TKey, TValue>(properties.KeyComparer, properties.ValueComparer); | ||
| modified = false; | ||
| foreach (var entry in properties) | ||
| { | ||
| if (canonicalProperties.TryGetValue(entry.Value, out var canonicalProperty)) | ||
| { | ||
| if (!ReferenceEquals(canonicalProperty, entry.Value)) | ||
| { | ||
| modified = true; | ||
| } | ||
|
|
||
| builder[entry.Key] = canonicalProperty; | ||
| } | ||
| else | ||
| { | ||
| canonicalProperties.Add(entry.Value, entry.Value); | ||
| builder[entry.Key] = entry.Value; | ||
| } | ||
| } | ||
|
|
||
| return modified ? builder.ToImmutable() : properties; | ||
| } |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes placement eligibility so a local silo that has transitioned to ShuttingDown is removed from compatible silo candidates even if membership still reports it active. GetCompatibleSilos now returns ImmutableArray and uses the cached active-silo list from ISiloStatusOracle.GetActiveSilos(), with active-silo filtering applied after placement filters so callers only receive active silos.