starred/windmill
1
0
Fork 0
mirror of https://github.com/windmill-labs/windmill.git synced 2026-08-21 00:02:23 +00:00
Code Issues Packages Projects Releases Wiki Activity

clarify state transition error

Browse Source
This commit is contained in:
Ruben Fiszel
2025-03-10 11:03:22 +01:00
parent e2897f6a41
commit 4a499cf5fc
1 changed files with 14 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/src/monitor.rs
+14 -3
View File
@@ -1917,16 +1917,27 @@ async fn handle_zombie_flows(db: &DB) -> error::Result<()> {
let id = flow.id.clone();
let last_ping = flow.last_ping.clone();
let now = now_from_db(db).await?;
let base_url = BASE_URL.read().await;
let workspace_id = flow.workspace_id.clone();
let reason = format!(
"{} was hanging in between 2 steps. Last ping: {last_ping:?} (now: {now})",
if flow.is_flow_step.unwrap_or(false) && flow.parent_job.is_some() {
format!("Flow was cancelled because subflow {id}")
format!("Flow was cancelled because subflow {id} ({base_url}/run/{id}?workspace={workspace_id})")
} else {
format!("Flow {id} was cancelled because it")
format!("Flow {id} ({base_url}/run/{id}?workspace={workspace_id}) was cancelled because it")
}
);
report_critical_error(reason.clone(), db.clone(), Some(&flow.workspace_id), None).await;
cancel_zombie_flow_job(db, flow.id, &flow.workspace_id, reason).await?;
cancel_zombie_flow_job(db, flow.id, &flow.workspace_id,
format!(r#"{reason}
This would happen if a worker was interrupted, killed or crashed while doing a state transition at the end of a job which is always an unexpected behavior that should never happen.
Please check your worker logs for more details and feel free to report it to the Windmill team on our Discord or support@windmill.dev (response for non EE customers will be best effort) with as much context as possible, ideally:
- Windmill version
- Worker logs right after the job referenced has finished running
- Is the error consistent when running the same flow
- A minimal flow and its flow.yaml that reproduces the error and that is importable in a fresh workspace
- Your infra setup (helm, docker-compose, configuration of the workers and their number, memory of the database, etc.)
"#)).await?;
}
}