k2safe/orbit
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add durable safe retries with uncertainty tracking and SDK examples

Browse Source
This commit is contained in:
k2safe
2026-09-17 15:43:44 +08:00
parent 9e3d4be5cc
commit d65cf96ea7
44 changed files with 1300 additions and 77 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
internal/storage/mysql/task_history.go
View File

@@ -91,11 +91,11 @@ func historyAttemptExists(ctx context.Context, tx *sql.Tx, ns, task, attempt cor
return err
}
const attemptHistoryColumns = `a.id,a.task_id,a.epoch,a.state,a.revision,a.lease_until,a.authorized_at,a.ended_at,COALESCE(a.error_code,''),a.created_at,a.updated_at`
const attemptHistoryColumns = `a.id,a.task_id,a.epoch,a.state,a.revision,a.lease_until,a.authorized_at,a.ended_at,COALESCE(a.error_code,''),a.retryable,a.created_at,a.updated_at`
func scanAttemptSummary(row interface{ Scan(...any) error }) (service.AttemptSummary, error) {
var a service.AttemptSummary
err := row.Scan(&a.ID, &a.TaskID, &a.Epoch, &a.State, &a.Revision, &a.LeaseUntil, &a.AuthorizedAt, &a.EndedAt, &a.ErrorCode, &a.CreatedAt, &a.UpdatedAt)
err := row.Scan(&a.ID, &a.TaskID, &a.Epoch, &a.State, &a.Revision, &a.LeaseUntil, &a.AuthorizedAt, &a.EndedAt, &a.ErrorCode, &a.Retryable, &a.CreatedAt, &a.UpdatedAt)
return a, err
}
func (s *TaskStore) ListTaskAttempts(ctx context.Context, t auth.Token, task core.ID, in service.HistoryPageInput) (service.AttemptPage, error) {
@@ -148,7 +148,7 @@ func (s *TaskStore) GetTaskAttempt(ctx context.Context, t auth.Token, task, atte
return err
}
var result []byte
err := tx.QueryRowContext(ctx, `SELECT `+attemptHistoryColumns+`,a.session_id,w.worker_instance_id,a.result FROM pulse_attempts a JOIN pulse_worker_sessions w ON w.namespace_id=a.namespace_id AND w.id=a.session_id WHERE a.namespace_id=? AND a.task_id=? AND a.id=?`, t.NamespaceID, task, attempt).Scan(&out.ID, &out.TaskID, &out.Epoch, &out.State, &out.Revision, &out.LeaseUntil, &out.AuthorizedAt, &out.EndedAt, &out.ErrorCode, &out.CreatedAt, &out.UpdatedAt, &out.SessionID, &out.WorkerInstanceID, &result)
err := tx.QueryRowContext(ctx, `SELECT `+attemptHistoryColumns+`,a.session_id,w.worker_instance_id,a.result FROM pulse_attempts a JOIN pulse_worker_sessions w ON w.namespace_id=a.namespace_id AND w.id=a.session_id WHERE a.namespace_id=? AND a.task_id=? AND a.id=?`, t.NamespaceID, task, attempt).Scan(&out.ID, &out.TaskID, &out.Epoch, &out.State, &out.Revision, &out.LeaseUntil, &out.AuthorizedAt, &out.EndedAt, &out.ErrorCode, &out.Retryable, &out.CreatedAt, &out.UpdatedAt, &out.SessionID, &out.WorkerInstanceID, &result)
if errors.Is(err, sql.ErrNoRows) {
return fault.NotFound
}
@@ -166,7 +166,7 @@ func (s *TaskStore) ListAttemptEvidence(ctx context.Context, t auth.Token, task,
if err := historyAttemptExists(ctx, tx, t.NamespaceID, task, attempt); err != nil {
return err
}
q := `SELECT id,attempt_id,outcome,COALESCE(error_code,''),reason,created_at FROM pulse_attempt_evidence WHERE namespace_id=? AND attempt_id=?`
q := `SELECT id,attempt_id,outcome,COALESCE(error_code,''),retryable,reason,created_at FROM pulse_attempt_evidence WHERE namespace_id=? AND attempt_id=?`
args := []any{t.NamespaceID, attempt}
if in.Cursor != "" {
c, err := decodeHistoryCursor(in.Cursor, t.NamespaceID, task, attempt)
@@ -185,7 +185,7 @@ func (s *TaskStore) ListAttemptEvidence(ctx context.Context, t auth.Token, task,
defer rows.Close()
for rows.Next() {
e := service.EvidenceSummary{TaskID: task, Disposition: "EVIDENCE_ONLY"}
if err := rows.Scan(&e.ID, &e.AttemptID, &e.Outcome, &e.ErrorCode, &e.Reason, &e.CreatedAt); err != nil {
if err := rows.Scan(&e.ID, &e.AttemptID, &e.Outcome, &e.ErrorCode, &e.Retryable, &e.Reason, &e.CreatedAt); err != nil {
return err
}
out.Items = append(out.Items, e)
@@ -213,7 +213,7 @@ func (s *TaskStore) GetAttemptEvidence(ctx context.Context, t auth.Token, task,
return err
}
var payload []byte
err := tx.QueryRowContext(ctx, `SELECT id,attempt_id,outcome,COALESCE(error_code,''),reason,created_at,message_id,payload FROM pulse_attempt_evidence WHERE namespace_id=? AND attempt_id=? AND id=?`, t.NamespaceID, attempt, evidence).Scan(&out.ID, &out.AttemptID, &out.Outcome, &out.ErrorCode, &out.Reason, &out.CreatedAt, &out.MessageID, &payload)
err := tx.QueryRowContext(ctx, `SELECT id,attempt_id,outcome,COALESCE(error_code,''),retryable,reason,created_at,message_id,payload FROM pulse_attempt_evidence WHERE namespace_id=? AND attempt_id=? AND id=?`, t.NamespaceID, attempt, evidence).Scan(&out.ID, &out.AttemptID, &out.Outcome, &out.ErrorCode, &out.Retryable, &out.Reason, &out.CreatedAt, &out.MessageID, &payload)
if errors.Is(err, sql.ErrNoRows) {
return fault.NotFound
}