From d4008e365503db283fd735553cca54d8a45eb03a Mon Sep 17 00:00:00 2001
From: Achim Rohn <achim@rohn.eu>
Date: Wed, 8 Apr 2026 15:41:22 +0200
Subject: [PATCH] Add methods to retry failed workflows and jobs across all
 steps and queues

---
 queue/general_queue.go | 25 ++++++++++++++++++++++---
 workflow/workflow.go   | 28 ++++++++++++++++++++++------
 2 files changed, 44 insertions(+), 9 deletions(-)

diff --git a/queue/general_queue.go b/queue/general_queue.go
index b1c6795..be65414 100644
--- a/queue/general_queue.go
+++ b/queue/general_queue.go
@@ -575,9 +575,8 @@ func (q *GeneralQueue) IncrementTries(ctx context.Context, jobID int, currentTri
 	return nil
 }
 
-// ResumeWorkflow resets jobs for a specific workflow in this queue so they can be processed again.
-// - All jobs with status InProgress are set back to Pending.
-// - All jobs with status Failed are set back to Pending and their NumberOfTries is reset to 0.
+// ResumeWorkflow resumes a specific workflow's execution in this queue by
+// setting all InProgress jobs back to Pending, allowing them to be picked up again.
 // After resetting, if the queue is marked running, it will be started.
 func (q *GeneralQueue) ResumeWorkflow(ctx context.Context, workflowId string) error {
 	// Reset in-progress jobs back to pending for the specific workflow
@@ -597,6 +596,26 @@ func (q *GeneralQueue) ResumeWorkflow(ctx context.Context, workflowId string) er
 		Debug("Reset ", inProgressCount, " in_progress jobs to pending for ", q.Name, " queue (workflow:", workflowId, ")")
 	}
 
+	// Start the queue again if it is configured as running
+	if isRunning, err := q.IsRunning(ctx); err != nil {
+		Error("Failed to check ", q.Name, " queue state:", err)
+	} else if isRunning {
+		Debug("restarting ", q.Name, " queue")
+		if err := q.Start(ctx); err != nil {
+			Error("Failed to restart ", q.Name, " queue:", err)
+			return err
+		}
+		Debug(q.Name, " queue auto-started")
+	} else {
+		Debug(q.Name, " queue not running")
+	}
+	return nil
+}
+
+// RetryFailedWorkflow retries failed jobs for a specific workflow in this queue by
+// setting all Failed jobs back to Pending and resetting their NumberOfTries to 0.
+// After resetting, if the queue is marked running, it will be started.
+func (q *GeneralQueue) RetryFailedWorkflow(ctx context.Context, workflowId string) error {
 	// Reset failed jobs back to pending and reset their try counter for the specific workflow
 	failedCount, err := q.client.GeneralQueue.Update().
 		Where(
diff --git a/workflow/workflow.go b/workflow/workflow.go
index db39dab..c377671 100644
--- a/workflow/workflow.go
+++ b/workflow/workflow.go
@@ -115,9 +115,9 @@ func (w *Workflow) GenerateId() string {
 	return w.Identifier + "_" + now.Format("20060102150405") + "_" + ersteller.RandomString(5)
 }
 
-// Resume restarts a specific workflow execution by its workflowId.
-// It resets failed and in-progress jobs of all steps back to pending and
-// starts their queues again if they are running.
+// Resume restarts a specific workflow execution by its workflowId by
+// setting any in-progress jobs of all steps back to pending, then starts
+// their queues again if they are running.
 func (w *Workflow) Resume(ctx context.Context, workflowId string) error {
 	var allErr error
 	for _, step := range w.AllSteps {
@@ -133,9 +133,8 @@ func (w *Workflow) Resume(ctx context.Context, workflowId string) error {
 }
 
 // ResumeAll restarts all executions of this workflow across all steps.
-// For each step it resets any in-progress jobs to pending and failed jobs to
-// pending with NumberOfTries reset to 0, then starts the queue again if it is
-// configured as running.
+// For each step it resets any in-progress jobs to pending, then starts the
+// queue again if it is configured as running.
 func (w *Workflow) ResumeAll(ctx context.Context) error {
 	var allErr error
 	for _, step := range w.AllSteps {
@@ -150,6 +149,23 @@ func (w *Workflow) ResumeAll(ctx context.Context) error {
 	return allErr
 }
 
+// RetryFailed restarts only failed jobs of a specific workflow execution by its
+// workflowId across all steps by resetting them to pending and their try count
+// to 0, then starts their queues again if they are running.
+func (w *Workflow) RetryFailed(ctx context.Context, workflowId string) error {
+	var allErr error
+	for _, step := range w.AllSteps {
+		if step.Queue == nil {
+			// Safety: ensure queues are initialized
+			step.initQueue()
+		}
+		if err := step.Queue.RetryFailedWorkflow(ctx, workflowId); err != nil {
+			allErr = errors.Join(allErr, fmt.Errorf("step %s: %w", step.Name, err))
+		}
+	}
+	return allErr
+}
+
 func NewCronTrigger(ctx context.Context, workflow *Workflow, d time.Duration) {
 	go func() {
 		ticker := time.NewTicker(d)