fix: bound cleanup goroutines, fix retry off-by-one, log dispatch errors in S3 adapter

[Copilot]

Three reliability issues in the S3 adapter identified during review of the add s3 adapter PR.

Changes

• retryS3Mutation off-by-one: attempt >= s3TxnRetryMaxAttempts with attempt starting at 0 allowed maxAttempts+1 total calls. Switched to for attempt := range s3TxnRetryMaxAttempts with an early break on the last iteration — exactly 8 attempts, no wasted backoff after final failure.

• Unbounded cleanupManifestBlobsAsync goroutines: High write/delete churn could spawn unbounded goroutines (each holding a 2m timeout context). Added a cleanupSem chan struct{} semaphore (capacity s3ManifestCleanupWorkers = 16) to S3Server; cleanup is skipped via non-blocking select when all slots are occupied rather than queuing indefinitely.

  select {
  case s.cleanupSem <- struct{}{}:
  default:
      // All cleanup workers busy; orphaned blobs persist until next overwrite/delete.
      return
  }
  go func() {
      defer func() { <-s.cleanupSem }()
      ...
  }()

• Silent coordinator.Dispatch failures in cleanupManifestBlobs: Dispatch errors were discarded with _, _. Now logged via slog.ErrorContext with structured fields (bucket, generation, object_key, upload_id, err) so cleanup failures are observable.

---

📍 Connect Copilot coding agent with Jira, Azure Boards or Linear to delegate work to Copilot in one click without leaving your project management tool.

[bootjp]

@copilot fix golangci-lint

[Copilot]

@copilot fix golangci-lint

Fixed in commit 52cf0ec. The cyclop linter reported cleanupManifestBlobs had complexity 11 (max 10). Extracted the per-part chunk key building into a new appendPartBlobKeys helper method, which reduces the complexity of cleanupManifestBlobs to within the limit. golangci-lint run ./adapter/... now reports 0 issues.