mocra-core 0.4.1

The mocra crawler framework runtime: errors, cache, utilities, domain models, downloader, data-plane queue, coordination, scheduler and engine.
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

use crate::common::state::State;
use crate::queue::compensation::Compensator;
#[cfg(feature = "store")]
use chrono::{DateTime, Utc};
use log::info;
#[cfg(feature = "store")]
use log::{error, warn};
#[cfg(feature = "store")]
use sea_orm::{ConnectionTrait, Statement, TransactionTrait};
use std::sync::Arc;
#[cfg(feature = "store")]
use tokio::time::{Duration, sleep};

/// Starts a periodic cleaner for stale `Running` tasks.
///
/// Every minute, the cleaner marks rows as `Failed` when `updated_at` is older
/// than `now - zombie_threshold_secs`.
pub async fn start_zombie_cleaner(
    state: Arc<State>,
    zombie_threshold_secs: i64,
    compensator: Option<Arc<dyn Compensator>>,
) {
    #[cfg(not(feature = "store"))]
    {
        let _ = (&state, zombie_threshold_secs, &compensator);
        info!("ZombieTaskCleaner disabled (no `store` feature); nothing to clean");
    }
    #[cfg(feature = "store")]
    {
        info!(
            "ZombieTaskCleaner started with threshold {}s",
            zombie_threshold_secs
        );

        loop {
            sleep(Duration::from_secs(60)).await;

            let threshold_time = Utc::now() - chrono::Duration::seconds(zombie_threshold_secs);

            match clean_zombies(&state, threshold_time, &compensator).await {
                Ok(count) => {
                    if count > 0 {
                        warn!("Cleaned up {} zombie tasks", count);
                    }
                }
                Err(e) => {
                    error!("Failed to clean zombie tasks: {}", e);
                }
            }
        }
    }
}

#[cfg(feature = "store")]
async fn clean_zombies(
    state: &State,
    threshold_time: DateTime<Utc>,
    compensator: &Option<Arc<dyn Compensator>>,
) -> Result<u64, sea_orm::DbErr> {
    let formatted_time = threshold_time.format("%Y-%m-%d %H:%M:%S").to_string();
    // No-DB (standalone) mode: there is no zombie task table to clean up.
    let Some(db) = state.db.as_ref() else {
        return Ok(0);
    };
    let backend = db.get_database_backend();

    let txn = db.begin().await?;

    // Mark zombie tasks as Failed inside the transaction.
    let update_sql = format!(
        "UPDATE base.task_result SET status = 'Failed', error = 'Zombie Task Detected (Timeout)' WHERE status = 'Running' AND updated_at < '{}'",
        formatted_time
    );
    let res = txn
        .execute(Statement::from_string(backend, update_sql))
        .await?;
    let affected = res.rows_affected();

    // Remove compensation records only for rows actually transitioned.
    if affected > 0 {
        if let Some(comp) = compensator {
            let select_sql = format!(
                "SELECT id FROM base.task_result WHERE status = 'Failed' AND error = 'Zombie Task Detected (Timeout)' AND updated_at < '{}'",
                formatted_time
            );
            if let Ok(rows) = txn
                .query_all(Statement::from_string(backend, select_sql))
                .await
            {
                for row in &rows {
                    if let Ok(id) = row.try_get::<String>("", "id") {
                        let _ = comp.remove_task("task", &id).await;
                    }
                }
            }
        }
    }

    txn.commit().await?;

    Ok(affected)
}