Claude finished @windmill-internal-app[bot]'s task —— View job • claude/issue-6408-20250819-0230 • [Create PR ➔](https://github.com/windmill-labs/windmill/compare/main...claude/issue-6408-20250819-0230?quick_pull=1&title=Issue%20%236408%3A%20Changes%20from%20Claude&body=This%20PR%20addresses%20issue%20%236408%0A%0AGenerated%20with%20%5BClaude%20Code%5D(https%3A%2F%2Fclaude.ai%2Fcode)

Adding Resilience to Pool Timeout in Flow Update ✅

Successfully implemented retry logic using backon for sqlx queries in update_flow_status_after_job_completion_internal function as requested by @rubenfiszel.

Changes Made:

Added backon imports to enable retry functionality
Created a retry_sqlx! macro that wraps sqlx operations with retry logic (3 attempts, 200ms delay)
Applied retry logic to critical queries:
- Initial flow status fetch query (most critical - could cause entire function to fail)
- Flow jobs results retrieval query (used for gathering job results)
Committed changes and created draft PR

Implementation Details:

File: /home/runner/work/windmill/windmill/backend/windmill-worker/src/worker_flow.rs

Key Changes:

Added retry macro (lines 72-90):

macro_rules! retry_sqlx {
    ($operation:expr) => {
        (|| async { $operation })
            .retry(
                ConstantBuilder::default()
                    .with_delay(std::time::Duration::from_millis(200))
                    .with_max_times(3)
                    .build(),
            )
            .notify(|err, dur| {
                tracing::warn!("Database operation failed, retrying in {dur:#?}, err: {err:#?}");
            })
            .sleep(tokio::time::sleep)
            .await
    };
}

Applied to critical queries:
- Flow status fetch (line 281) - Prevents function failure on pool timeout
- Job results retrieval (line 1468) - Ensures job results are properly collected

Why this approach:

Focused on queries outside transactions (safer to retry)
Targets most critical queries that would cause workflow failures
Uses proven retry patterns from existing codebase
Minimal invasive changes with maximum resilience benefit

Pull Request Created 🚀

Draft PR: #6412 - feat: add resilience to pool timeout in flow update

The implementation provides resilience to database pool timeouts in the most critical parts of the flow update process, significantly reducing the risk of workflow failures due to temporary database connectivity issues.

Status: Complete ✅

All requested changes have been implemented and are ready for review in the draft pull request.

more resilience to pool timeout in flow update #6408

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions