starred/windmill
1
0
Fork 0
mirror of https://github.com/windmill-labs/windmill.git synced 2026-10-23 08:03:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: stop holding a pool connection while waiting for another (#11902)

Browse Source 
* fix: stop holding a pool connection while waiting for another

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix: count only earlier occurrences as schedule handler history

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* chore: update ee-repo-ref to ac53164fd27fe962a020f3ee944949df1f760603

This commit updates the EE repository reference after PR #904 was merged in windmill-ee-private.

Previous ee-repo-ref: 047764f902f10fce0156138c4240ba5d1a2e969a

New ee-repo-ref: ac53164fd27fe962a020f3ee944949df1f760603

Automated by sync-ee-ref workflow.

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: windmill-internal-app[bot] <windmill-internal-app[bot]@users.noreply.github.com>
This commit is contained in:
Ruben Fiszel
2026-10-20 19:11:46 +02:00
committed by GitHub
co-authored by Claude Opus 5.5 windmill-internal-app[bot]
parent 613adb5287
commit 2c559b452f
8 changed files with 266 additions and 52 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/windmill-queue/src/jobs.rs
+48 -41
View File
@@ -1641,6 +1641,7 @@ async fn commit_completed_job<T: Serialize + Send + Sync + ValidableJson>(
}
let mut _skip_downstream_error_handlers = false;
let mut schedule_for_handlers = None;
if completed_job.is_flow_step() {
if let Some(parent_job) = monitor_parent {
let r = sqlx::query_scalar!(
@@ -1720,49 +1721,9 @@ async fn commit_completed_job<T: Serialize + Send + Sync + ValidableJson>(
// per-occurrence failure/recovery status across the whole retry
// chain, so multi-count/exact handler policies work even though
// each attempt is its own completed job.
#[cfg(all(feature = "enterprise", feature = "private"))]
if !retry_pending {
if let Err(err) = crate::jobs_ee::apply_schedule_handlers(
db,
&schedule,
&script_path,
&completed_job.workspace_id,
success,
result,
job_id,
// Current occurrence's root: the terminal native-retry
// attempt's parent, else the job itself.
completed_job.parent_job.unwrap_or(job_id),
completed_job.started_at.unwrap_or(chrono::Utc::now()),
completed_job.priority,
)
.warn_after_seconds(10)
.await
{
if !success {
tracing::error!("Could not apply schedule error handler: {}", err);
let base_url = windmill_common::BASE_URL.load();
let w_id: &String = &completed_job.workspace_id;
if !matches!(err, Error::QuotaExceeded(_)) {
report_error_to_workspace_handler_or_critical_side_channel(
&completed_job,
db,
format!(
"Failed to push schedule error handler job to handle failed job ({base_url}/run/{}?workspace={w_id}): {}",
completed_job.id,
err
),
)
.warn_after_seconds(10)
.await;
}
} else {
tracing::error!("Could not apply schedule recovery handler: {}", err);
}
};
schedule_for_handlers = Some((schedule, script_path));
}
#[cfg(not(all(feature = "enterprise", feature = "private")))]
let _ = retry_pending;
} else {
tracing::error!(
"Schedule {schedule_path} in {} not found. Impossible to schedule again and apply schedule handlers",
@@ -1824,6 +1785,52 @@ async fn commit_completed_job<T: Serialize + Send + Sync + ValidableJson>(
tx.commit().warn_after_seconds(10).await?;
// The handlers read and push through `db`, so they run after the commit: inside the
// transaction they would hold its connection while waiting for more from the same pool.
#[cfg(all(feature = "enterprise", feature = "private"))]
if let Some((schedule, script_path)) = schedule_for_handlers {
if let Err(err) = crate::jobs_ee::apply_schedule_handlers(
db,
&schedule,
script_path,
&completed_job.workspace_id,
success,
result,
job_id,
// Current occurrence's root: the terminal native-retry
// attempt's parent, else the job itself.
completed_job.parent_job.unwrap_or(job_id),
completed_job.started_at.unwrap_or(chrono::Utc::now()),
completed_job.priority,
)
.warn_after_seconds(10)
.await
{
if !success {
tracing::error!("Could not apply schedule error handler: {}", err);
let base_url = windmill_common::BASE_URL.load();
let w_id: &String = &completed_job.workspace_id;
if !matches!(err, Error::QuotaExceeded(_)) {
report_error_to_workspace_handler_or_critical_side_channel(
&completed_job,
db,
format!(
"Failed to push schedule error handler job to handle failed job ({base_url}/run/{}?workspace={w_id}): {}",
completed_job.id,
err
),
)
.warn_after_seconds(10)
.await;
}
} else {
tracing::error!("Could not apply schedule recovery handler: {}", err);
}
};
}
#[cfg(not(all(feature = "enterprise", feature = "private")))]
let _ = schedule_for_handlers;
log_completed_job(completed_job, completed.duration_ms, success);
// tracing::info!("completed job: {:?}", start.elapsed().as_micros());
Ok((
backend/windmill-queue/tests/native_retry_test.rs
+3 -2
View File
@@ -339,7 +339,7 @@ mod native_retry {
// -> failure: handler child must NOT count as a recovery
// f: root fails, only a WAC inline child (unmarked) succeeds
// -> failure: WAC inline child must NOT count as a recovery
// d: the current occurrence (excluded by `j.id != $4`)
// d: the current occurrence (excluded with any later one by the created_at bound)
let (a, b, c, e, f, d) = (
Uuid::new_v4(),
Uuid::new_v4(),
@@ -371,7 +371,8 @@ mod native_retry {
))
FROM v2_job j JOIN v2_job_completed USING (id)
WHERE j.workspace_id = $1 AND trigger_kind = 'schedule' AND trigger = $2
AND parent_job IS NULL AND runnable_path = $3 AND j.id != $4
AND parent_job IS NULL AND runnable_path = $3
AND j.created_at < (SELECT created_at FROM v2_job WHERE id = $4)
ORDER BY created_at DESC",
)
.bind(WS)