[DB-2199] Fix write timeout regressions introduced with multi-stream appends

[shaan1337]

No description provided.

[linear-code]

DB-2199

[qodo-code-review]

PR Summary by Qodo

Fix write-timeout handling for multi-stream appends (checkpoint + parked writes)

🐞 Bug fix 🧪 Tests 🕐 20-40 Minutes

AI Description

• Guard event-number reads on successful writes to avoid exceptions on timeouts.
• Ensure checkpoint writes retry properly and fail projections after retry exhaustion.
• Ensure park-message timeouts complete callbacks without updating parked-message stats.

Diagram

graph TD
  PSMP["PersistentSubscriptionMessageParker"] -->|"writes parked event"| IOD["IODispatcher"] --> WEC["WriteEventsCompleted"] -->|"callback + result"| PSMP
  CPCW["CoreProjectionCheckpointWriter"] -->|"write checkpoint"| IOD --> WEC -->|"result + event#"| CPCW
  CPCW -->|"schedule retry"| TIMER["TimerService"] --> CPCW
  CPCW -->|"on exhaustion"| PROJ["Projection failure"]
  subgraph Legend
    direction LR
    _cmp([Component]) ~~~ _msg["Message"]
  end

Loading

High-Level Assessment

The following are alternative approaches to this PR:

1. Populate First/LastEventNumbers on failures in WriteEventsCompleted

• ➕ Prevents this entire class of bugs across all call sites
• ➕ Simplifies downstream callback code (no success-guard needed)
• ➖ Risky semantic change: failure completions may not have meaningful event numbers
• ➖ Broader blast radius; may mask assumptions elsewhere

2. Introduce safe accessors (e.g., TryGetSingleEventNumber)

• ➕ Centralizes the invariant and avoids repeated result checks
• ➕ Makes failure-path handling explicit and self-documenting
• ➖ Requires API changes or extensions across multiple projects
• ➖ Still needs callers to decide fallback behavior

3. Always pass EventNumber.Invalid from dispatcher on non-success

• ➕ Keeps message shape unchanged while standardizing failure behavior
• ➕ Localizes change to the dispatch boundary
• ➖ May hide cases where event numbers are unexpectedly missing on success
• ➖ Still requires downstream to interpret Invalid correctly

Recommendation: The PR’s approach (guard reads on OperationResult.Success and add regression tests) is the safest, lowest-blast-radius fix for the regression. Consider a follow-up to add a standardized safe accessor or dispatcher-level normalization to reduce future repeated footguns, but avoid changing WriteEventsCompleted failure semantics unless you can verify all downstream assumptions.

Files changed (4) +95 / -4

Bug fix (2) +9 / -4

PersistentSubscriptionMessageParker.csOnly update parked-message state on successful writes +5/-3

Only update parked-message state on successful writes

• Guards reads of LastEventNumbers.Single behind OperationResult.Success to avoid exceptions when write metadata is absent on failures. Ensures failures still invoke the completion callback without mutating parked-message counters/timestamps.

src/KurrentDB.Core/Services/PersistentSubscription/PersistentSubscriptionMessageParker.cs

CoreProjectionCheckpointWriter.csAvoid reading FirstEventNumbers on failed checkpoint writes +4/-1

Avoid reading FirstEventNumbers on failed checkpoint writes

• Updates the write completion callback to only read FirstEventNumbers.Single on success, otherwise passing EventNumber.Invalid. Prevents failure-path exceptions that previously bypassed retry/failure handling.

src/KurrentDB.Projections.V1/Services/Processing/Checkpointing/CoreProjectionCheckpointWriter.cs

Tests (2) +86 / -0

PersistentSubscriptionMessageParkerTests.csAdd regression test for park write commit-timeout behavior +27/-0

Add regression test for park write commit-timeout behavior

• Adds a fixture that forces parked-message writes to time out and asserts the completion callback runs with the failure result. Verifies no parked-message stats are updated when the write fails.

src/KurrentDB.Core.Tests/Services/PersistentSubscription/PersistentSubscriptionMessageParkerTests.cs

when_the_checkpoint_write_times_out.csAdd regression test for checkpoint write timeouts and retry exhaustion +59/-0

Add regression test for checkpoint write timeouts and retry exhaustion

• Introduces a test that forces checkpoint writes to return CommitTimeout repeatedly. Confirms retries are attempted until the scheduler stops and the projection is failed after exhausting retry limits.

src/KurrentDB.Projections.V1.Tests/Services/core_projection/checkpoint_manager/when_the_checkpoint_write_times_out.cs

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

Great, no issues found!

Qodo reviewed your code and found no material issues that require review

_{To customize comments, go to the Qodo configuration screen, or learn more in the docs.}

[github-actions]

🚨 @shaan1337 Failed to create cherry Pick PR due to error:

RequestError [HttpError]: Merge conflict
   at /home/runner/work/_actions/kurrent-io/Automations/master/cherry-pick-pr-for-label/node_modules/@octokit/request/dist-node/index.js:66:23
   at process.processTicksAndRejections (node:internal/process/task_queues:104:5) {
 status: '409',
 headers: {
   'access-control-allow-origin': '*',
   'access-control-expose-headers': 'ETag, Link, Location, Retry-After, X-GitHub-OTP, X-RateLimit-Limit, X-RateLimit-Remaining, X-RateLimit-Used, X-RateLimit-Resource, X-RateLimit-Reset, X-OAuth-Scopes, X-Accepted-OAuth-Scopes, X-Poll-Interval, X-GitHub-Media-Type, X-GitHub-SSO, X-GitHub-Request-Id, Deprecation, Sunset, Warning',
   'content-length': '127',
   'content-security-policy': "default-src 'none'",
   'content-type': 'application/json; charset=utf-8',
   date: 'Wed, 05 Aug 2026 10:03:31 GMT',
   'referrer-policy': 'origin-when-cross-origin, strict-origin-when-cross-origin',
   server: 'github.com',
   'strict-transport-security': 'max-age=31536000; includeSubdomains; preload',
   vary: 'Accept-Encoding, Accept, X-Requested-With',
   'x-accepted-github-permissions': 'contents=write',
   'x-content-type-options': 'nosniff',
   'x-frame-options': 'deny',
   'x-github-api-version-selected': '2022-11-28',
   'x-github-edge-region': 'iad',
   'x-github-media-type': 'github.v3; format=json',
   'x-github-request-id': 'E7C1:2F4E58:1601F3C:49FD2FD:6A730A73',
   'x-ratelimit-limit': '5000',
   'x-ratelimit-remaining': '4995',
   'x-ratelimit-reset': '1785927808',
   'x-ratelimit-resource': 'core',
   'x-ratelimit-used': '5',
   'x-xss-protection': '0'
 },
 request: {
   method: 'POST',
   url: 'https://api.github.com/repos/kurrent-io/KurrentDB/merges',
   headers: {
     accept: 'application/vnd.github.v3+json',
     'user-agent': 'octokit-core.js/3.3.2 Node.js/24',
     authorization: 'bearer [REDACTED]',
     'content-type': 'application/json; charset=utf-8'
   },
   body: '{"base":"cherry-pick-cherry-pick/5699/shaan1337/fix-write-timeout-msa-bugs-release/v26.0-087cd9af-0a53-419c-9436-3dfa5dc0b62c","commit_message":"Merge d6c2f3b00a8906686d84f9b5eb313edb02436d27 into cherry-pick-cherry-pick/5699/shaan1337/fix-write-timeout-msa-bugs-release/v26.0-087cd9af-0a53-419c-9436-3dfa5dc0b62c [skip ci]\\n\\n\\nskip-checks: true\\n","head":"d6c2f3b00a8906686d84f9b5eb313edb02436d27"}',
   request: { agent: [Agent], hook: [Function: bound bound register] }
 },
 documentation_url: 'https://docs.github.com/rest/branches/branches#merge-a-branch'
}

🚨👉 Check https://github.com/kurrent-io/KurrentDB/actions/runs/30995845342

[github-actions]

@shaan1337 👉 Created pull request targeting release/v26.1: #5700