cloacina 0.3.2

A Rust library for resilient task execution and orchestration.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
/*
 *  Copyright 2025-2026 Colliery Software
 *
 *  Licensed under the Apache License, Version 2.0 (the "License");
 *  you may not use this file except in compliance with the License.
 *  You may obtain a copy of the License at
 *
 *      http://www.apache.org/licenses/LICENSE-2.0
 *
 *  Unless required by applicable law or agreed to in writing, software
 *  distributed under the License is distributed on an "AS IS" BASIS,
 *  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 *  See the License for the specific language governing permissions and
 *  limitations under the License.
 */

//! Task claiming and retry scheduling operations.
//!
//! All operations are transactional: state changes and execution events
//! are written atomically. If either fails, both are rolled back.

use super::{ClaimResult, TaskExecutionDAL};
use crate::dal::unified::models::{NewUnifiedExecutionEvent, UnifiedTaskExecution};
use crate::database::schema::unified::{execution_events, task_executions, task_outbox};
use crate::database::universal_types::{UniversalTimestamp, UniversalUuid};
use crate::error::ValidationError;
use crate::models::execution_event::ExecutionEventType;
use crate::models::task_execution::TaskExecution;
use diesel::prelude::*;
use uuid::Uuid;

impl<'a> TaskExecutionDAL<'a> {
    /// Updates a task's retry schedule with a new attempt count and retry time.
    ///
    /// This operation is transactional: the status update and execution event
    /// are written atomically.
    pub async fn schedule_retry(
        &self,
        task_id: UniversalUuid,
        retry_at: UniversalTimestamp,
        new_attempt: i32,
    ) -> Result<(), ValidationError> {
        crate::dispatch_backend!(
            self.dal.backend(),
            self.schedule_retry_postgres(task_id, retry_at, new_attempt)
                .await,
            self.schedule_retry_sqlite(task_id, retry_at, new_attempt)
                .await
        )
    }

    #[cfg(feature = "postgres")]
    async fn schedule_retry_postgres(
        &self,
        task_id: UniversalUuid,
        retry_at: UniversalTimestamp,
        new_attempt: i32,
    ) -> Result<(), ValidationError> {
        use crate::dal::unified::models::NewUnifiedTaskOutbox;
        use diesel::connection::Connection;

        let conn = self
            .dal
            .database
            .get_postgres_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        conn.interact(move |conn| {
            conn.transaction::<_, diesel::result::Error, _>(|conn| {
                let now = UniversalTimestamp::now();

                // Get task info for event
                let task: UnifiedTaskExecution =
                    task_executions::table.find(task_id).first(conn)?;

                // Update task retry state
                diesel::update(task_executions::table.find(task_id))
                    .set((
                        task_executions::status.eq("Ready"),
                        task_executions::attempt.eq(new_attempt),
                        task_executions::retry_at.eq(Some(retry_at)),
                        task_executions::started_at.eq(None::<UniversalTimestamp>),
                        task_executions::completed_at.eq(None::<UniversalTimestamp>),
                        task_executions::updated_at.eq(now),
                    ))
                    .execute(conn)?;

                // Insert execution event with retry details
                let event_data = serde_json::json!({
                    "attempt": new_attempt,
                    "retry_at": retry_at.to_string()
                })
                .to_string();
                let event = NewUnifiedExecutionEvent {
                    id: UniversalUuid::new_v4(),
                    pipeline_execution_id: task.pipeline_execution_id,
                    task_execution_id: Some(task_id),
                    event_type: ExecutionEventType::TaskRetryScheduled.as_str().to_string(),
                    event_data: Some(event_data),
                    worker_id: None,
                    created_at: now,
                };
                diesel::insert_into(execution_events::table)
                    .values(&event)
                    .execute(conn)?;

                // Insert outbox entry for work distribution
                // Use retry_at as created_at so workers won't claim until retry time
                let outbox_entry = NewUnifiedTaskOutbox {
                    task_execution_id: task_id,
                    created_at: retry_at,
                };
                diesel::insert_into(task_outbox::table)
                    .values(&outbox_entry)
                    .execute(conn)?;

                Ok(())
            })
        })
        .await
        .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(())
    }

    #[cfg(feature = "sqlite")]
    async fn schedule_retry_sqlite(
        &self,
        task_id: UniversalUuid,
        retry_at: UniversalTimestamp,
        new_attempt: i32,
    ) -> Result<(), ValidationError> {
        use crate::dal::unified::models::NewUnifiedTaskOutbox;
        use diesel::connection::Connection;

        let conn = self
            .dal
            .database
            .get_sqlite_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        conn.interact(move |conn| {
            conn.transaction::<_, diesel::result::Error, _>(|conn| {
                let now = UniversalTimestamp::now();

                // Get task info for event
                let task: UnifiedTaskExecution =
                    task_executions::table.find(task_id).first(conn)?;

                // Update task retry state
                diesel::update(task_executions::table.find(task_id))
                    .set((
                        task_executions::status.eq("Ready"),
                        task_executions::attempt.eq(new_attempt),
                        task_executions::retry_at.eq(Some(retry_at)),
                        task_executions::started_at.eq(None::<UniversalTimestamp>),
                        task_executions::completed_at.eq(None::<UniversalTimestamp>),
                        task_executions::updated_at.eq(now),
                    ))
                    .execute(conn)?;

                // Insert execution event with retry details
                let event_data = serde_json::json!({
                    "attempt": new_attempt,
                    "retry_at": retry_at.to_string()
                })
                .to_string();
                let event = NewUnifiedExecutionEvent {
                    id: UniversalUuid::new_v4(),
                    pipeline_execution_id: task.pipeline_execution_id,
                    task_execution_id: Some(task_id),
                    event_type: ExecutionEventType::TaskRetryScheduled.as_str().to_string(),
                    event_data: Some(event_data),
                    worker_id: None,
                    created_at: now,
                };
                diesel::insert_into(execution_events::table)
                    .values(&event)
                    .execute(conn)?;

                // Insert outbox entry for work distribution
                // Use retry_at as created_at so workers won't claim until retry time
                let outbox_entry = NewUnifiedTaskOutbox {
                    task_execution_id: task_id,
                    created_at: retry_at,
                };
                diesel::insert_into(task_outbox::table)
                    .values(&outbox_entry)
                    .execute(conn)?;

                Ok(())
            })
        })
        .await
        .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(())
    }

    /// Atomically claims up to `limit` ready tasks for execution.
    ///
    /// This operation is transactional: the status update and execution events
    /// are written atomically for all claimed tasks.
    pub async fn claim_ready_task(
        &self,
        limit: usize,
    ) -> Result<Vec<ClaimResult>, ValidationError> {
        crate::dispatch_backend!(
            self.dal.backend(),
            self.claim_ready_task_postgres(limit).await,
            self.claim_ready_task_sqlite(limit).await
        )
    }

    #[cfg(feature = "postgres")]
    async fn claim_ready_task_postgres(
        &self,
        limit: usize,
    ) -> Result<Vec<ClaimResult>, ValidationError> {
        use diesel::connection::Connection;

        let conn = self
            .dal
            .database
            .get_postgres_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        let limit = limit as i64;

        #[derive(Debug, QueryableByName, Clone)]
        #[diesel(check_for_backend(diesel::pg::Pg))]
        struct PgClaimResult {
            #[diesel(sql_type = diesel::sql_types::Uuid)]
            id: Uuid,
            #[diesel(sql_type = diesel::sql_types::Uuid)]
            pipeline_execution_id: Uuid,
            #[diesel(sql_type = diesel::sql_types::Text)]
            task_name: String,
            #[diesel(sql_type = diesel::sql_types::Integer)]
            attempt: i32,
        }

        let pg_results: Vec<PgClaimResult> = conn
            .interact(move |conn| {
                conn.transaction::<_, diesel::result::Error, _>(|conn| {
                    let now = UniversalTimestamp::now();

                    // Claim tasks from outbox with FOR UPDATE SKIP LOCKED:
                    // 1. Select outbox entries with lock (skip locked rows)
                    //    - Filter by created_at <= NOW() to respect retry delays
                    // 2. Delete those outbox entries
                    // 3. Update corresponding task_executions to Running
                    // 4. Return task details
                    let claimed: Vec<PgClaimResult> = diesel::sql_query(format!(
                        r#"
                        WITH claimed_outbox AS (
                            DELETE FROM task_outbox
                            WHERE id IN (
                                SELECT id FROM task_outbox
                                WHERE created_at <= NOW()
                                ORDER BY created_at ASC
                                LIMIT {}
                                FOR UPDATE SKIP LOCKED
                            )
                            RETURNING task_execution_id
                        )
                        UPDATE task_executions
                        SET status = 'Running', started_at = NOW(), updated_at = NOW()
                        FROM claimed_outbox
                        WHERE task_executions.id = claimed_outbox.task_execution_id
                        RETURNING task_executions.id, task_executions.pipeline_execution_id, task_executions.task_name, task_executions.attempt
                        "#,
                        limit
                    ))
                    .load(conn)?;

                    // Insert execution events for all claimed tasks
                    for task in &claimed {
                        let event = NewUnifiedExecutionEvent {
                            id: UniversalUuid::new_v4(),
                            pipeline_execution_id: UniversalUuid(task.pipeline_execution_id),
                            task_execution_id: Some(UniversalUuid(task.id)),
                            event_type: ExecutionEventType::TaskClaimed.as_str().to_string(),
                            event_data: None,
                            worker_id: None,
                            created_at: now,
                        };
                        diesel::insert_into(execution_events::table)
                            .values(&event)
                            .execute(conn)?;
                    }

                    Ok(claimed)
                })
            })
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(pg_results
            .into_iter()
            .map(|pg| ClaimResult {
                id: UniversalUuid(pg.id),
                pipeline_execution_id: UniversalUuid(pg.pipeline_execution_id),
                task_name: pg.task_name,
                attempt: pg.attempt,
            })
            .collect())
    }

    #[cfg(feature = "sqlite")]
    async fn claim_ready_task_sqlite(
        &self,
        limit: usize,
    ) -> Result<Vec<ClaimResult>, ValidationError> {
        use crate::dal::unified::models::UnifiedTaskOutbox;
        use diesel::connection::Connection;

        let conn = self
            .dal
            .database
            .get_sqlite_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        let limit = limit as i64;

        // SQLite doesn't support FOR UPDATE SKIP LOCKED, so we use an IMMEDIATE transaction
        // to acquire a write lock at the start, preventing race conditions between workers.
        // This serializes concurrent claim attempts, ensuring each task is claimed exactly once.
        let tasks: Vec<UnifiedTaskExecution> = conn
            .interact(
                move |conn| -> Result<Vec<UnifiedTaskExecution>, diesel::result::Error> {
                    // Use IMMEDIATE transaction to acquire write lock immediately
                    // This prevents TOCTOU race conditions between SELECT and UPDATE
                    conn.transaction::<Vec<UnifiedTaskExecution>, diesel::result::Error, _>(
                        |conn| {
                            let now = UniversalTimestamp::now();

                            // Select oldest outbox entries within the transaction
                            // Filter by created_at <= NOW() to respect retry delays
                            let outbox_entries: Vec<UnifiedTaskOutbox> = task_outbox::table
                                .filter(task_outbox::created_at.le(now))
                                .order(task_outbox::created_at.asc())
                                .limit(limit)
                                .load(conn)?;

                            if outbox_entries.is_empty() {
                                return Ok(Vec::new());
                            }

                            // Collect task execution IDs and outbox IDs
                            let task_ids: Vec<_> =
                                outbox_entries.iter().map(|o| o.task_execution_id).collect();
                            let outbox_ids: Vec<_> = outbox_entries.iter().map(|o| o.id).collect();

                            // Delete outbox entries
                            diesel::delete(task_outbox::table)
                                .filter(task_outbox::id.eq_any(&outbox_ids))
                                .execute(conn)?;

                            // Load task executions for the claimed tasks
                            let claimed_tasks: Vec<UnifiedTaskExecution> = task_executions::table
                                .filter(task_executions::id.eq_any(&task_ids))
                                .load(conn)?;

                            // Batch update all tasks to Running in a single query
                            diesel::update(task_executions::table)
                                .filter(task_executions::id.eq_any(&task_ids))
                                .set((
                                    task_executions::status.eq("Running"),
                                    task_executions::started_at.eq(Some(now)),
                                    task_executions::updated_at.eq(now),
                                ))
                                .execute(conn)?;

                            // Insert execution events for all claimed tasks
                            for task in &claimed_tasks {
                                let event = NewUnifiedExecutionEvent {
                                    id: UniversalUuid::new_v4(),
                                    pipeline_execution_id: task.pipeline_execution_id,
                                    task_execution_id: Some(task.id),
                                    event_type: ExecutionEventType::TaskClaimed
                                        .as_str()
                                        .to_string(),
                                    event_data: None,
                                    worker_id: None,
                                    created_at: now,
                                };
                                diesel::insert_into(execution_events::table)
                                    .values(&event)
                                    .execute(conn)?;
                            }

                            Ok(claimed_tasks)
                        },
                    )
                },
            )
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(tasks
            .into_iter()
            .map(|task| ClaimResult {
                id: task.id,
                pipeline_execution_id: task.pipeline_execution_id,
                task_name: task.task_name,
                attempt: task.attempt,
            })
            .collect())
    }

    /// Retrieves tasks that are ready for retry (retry_at time has passed).
    pub async fn get_ready_for_retry(&self) -> Result<Vec<TaskExecution>, ValidationError> {
        crate::dispatch_backend!(
            self.dal.backend(),
            self.get_ready_for_retry_postgres().await,
            self.get_ready_for_retry_sqlite().await
        )
    }

    #[cfg(feature = "postgres")]
    async fn get_ready_for_retry_postgres(&self) -> Result<Vec<TaskExecution>, ValidationError> {
        let conn = self
            .dal
            .database
            .get_postgres_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        let now = UniversalTimestamp::now();
        let ready_tasks: Vec<UnifiedTaskExecution> = conn
            .interact(move |conn| {
                task_executions::table
                    .filter(task_executions::status.eq("Ready"))
                    .filter(
                        task_executions::retry_at
                            .is_null()
                            .or(task_executions::retry_at.le(now)),
                    )
                    .load(conn)
            })
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(ready_tasks.into_iter().map(Into::into).collect())
    }

    #[cfg(feature = "sqlite")]
    async fn get_ready_for_retry_sqlite(&self) -> Result<Vec<TaskExecution>, ValidationError> {
        let conn = self
            .dal
            .database
            .get_sqlite_connection()
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))?;

        let now = UniversalTimestamp::now();
        let ready_tasks: Vec<UnifiedTaskExecution> = conn
            .interact(move |conn| {
                task_executions::table
                    .filter(task_executions::status.eq("Ready"))
                    .filter(
                        task_executions::retry_at
                            .is_null()
                            .or(task_executions::retry_at.le(now)),
                    )
                    .load(conn)
            })
            .await
            .map_err(|e| ValidationError::ConnectionPool(e.to_string()))??;

        Ok(ready_tasks.into_iter().map(Into::into).collect())
    }
}