-
Notifications
You must be signed in to change notification settings - Fork 62
Jhwang/add ability to modify partition #448
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
On success, it should redirect back to the Backfill Show page. Alternatively it'd be nice to have the change form inline like the other config change buttons which turn into input boxes. |
Changes are done but let me see if I can add some tests. |
This change introduces a Cancel button to the UI for backfills that are currently paused. Users can now manually transition a paused backfill to a CANCELLED state, providing better control over workflows that are no longer needed. Changes Include: - Added a cancel button in the backfill details view (only shown for paused backfills). - Updated run partition table state column and backfillruns table state column, handling to correctly mark backfills as CANCELLED when triggered. Next Pr will create a new deleted column to support hiding the backfill that are soft deleted. https://github.com/user-attachments/assets/b0072300-d0ef-47cb-946e-b52f97e97073 
## Problem The DynamoDB BatchWriteItem API often has transient failures with unprocessed items that cause entire Backfila batches to fail. These could be more granularly retried within a run to avoid Backfila getting stuck on a batch. ## Solution Improve the retry mechanism to better align with [AWS BatchWriteItem best practices](https://docs.aws.amazon.com/amazondynamodb/latest/APIReference/API_BatchWriteItem.html): - Collect unprocessed items across all batches and retry them together - Use exponential backoff with jitter to handle throttling - Only count towards retry limit when no progress is made - Provide more detailed error reporting
…#452) ## Problem The DynamoDB BatchWriteItem implementation in UpdateInPlaceDynamoDbBackfill currently lacks handling for ApiCallTimeoutException. When these timeouts occur, the entire batch fails without any retry attempts, causing backfills to fail unnecessarily. ## Solution Add comprehensive timeout handling with these features: - Track and retry chunks that experience API timeouts - Use exponential backoff with jitter for retries - Only increment the timeout counter when all chunks in an iteration timeout - Reset the timeout counter if any chunk succeeds - Maintain separate counters for timeouts vs unprocessed items - Provide detailed error context through suppressed exceptions The implementation is designed to be resilient to transient timeouts while still protecting against systemic failures. It coordinates the backoff strategy between timeout retries and unprocessed item retries.
Enabe dynamic partition adjustments for paused backfills
Screen.Recording.2025-05-07.at.12.16.07.PM.mov