Skip to content

[PECOBLR-3984] Add jitter to chunk download retries - #1647

Open
prathamesh-db wants to merge 3 commits into
databricks:mainfrom
prathamesh-db:peco-ai-PECOBLR-3984
Open

[PECOBLR-3984] Add jitter to chunk download retries#1647
prathamesh-db wants to merge 3 commits into
databricks:mainfrom
prathamesh-db:peco-ai-PECOBLR-3984

Conversation

@prathamesh-db

@prathamesh-db prathamesh-db commented Aug 20, 2026

Copy link
Copy Markdown
Contributor

Description

Adds equal jitter (750–1,500 ms) to the existing chunk-download retry delay in both batch and streaming paths. This reduces synchronized retry bursts while preserving five attempts and the existing six-second maximum total retry wait.

Jira: https://databricks.atlassian.net/browse/PECOBLR-3984

Testing

  • Added deterministic coverage for both jitter boundaries.
  • Updated batch and streaming tests to verify the selected nonzero delays reach the sleep path.
  • Spotless passes; the focused Maven suite passes all 10 tests.

Telemetry Errors

  • Not applicable — this PR does not add or change a telemetry-visible error.
  • Applicable — the error uses DatabricksDriverErrorCode where appropriate, and any new code is uniquely numbered and tested.
  • Applicable — its driver/server/user classification is linked, or maintainer help is requested because the author cannot access the classification.

Additional Notes to the Reviewer

The retry count and existing CHUNK_DOWNLOAD_ERROR telemetry code are unchanged. This mitigation reduces correlated retries; it does not claim to eliminate every chunk-download failure.

This draft PR was generated by the PECO Agent Workflow and then narrowed to the telemetry-backed mitigation above.

Signed-off-by: Prathamesh Baviskar <prathamesh.baviskar@databricks.com>
Signed-off-by: Prathamesh Baviskar <prathamesh.baviskar@databricks.com>
@prathamesh-db prathamesh-db changed the title [PECO-AI] PECOBLR-3984: [driver-error] Investigate CHUNK_DOWNLOAD_ERROR [PECOBLR-3984] Add jitter to chunk download retries Aug 20, 2026
@prathamesh-db
prathamesh-db marked this pull request as ready for review August 21, 2026 05:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant