datafusion-dist 0.1.0

A distributed streaming execution library for Apache DataFusion
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

use std::{
    collections::{HashMap, HashSet},
    fmt::Debug,
    sync::Arc,
};

use datafusion::physical_plan::ExecutionPlan;
use serde::{Deserialize, Serialize};
use uuid::Uuid;

use crate::{
    DistResult, RecordBatchStream,
    cluster::NodeId,
    planner::{StageId, TaskId},
    runtime::{StageState, TaskMetrics, TaskSet},
};

#[async_trait::async_trait]
pub trait DistNetwork: Debug + Send + Sync {
    fn local_node(&self) -> NodeId;

    // Send task plan
    async fn send_tasks(&self, node_id: NodeId, scheduled_tasks: ScheduledTasks) -> DistResult<()>;

    // Execute task plan
    async fn execute_task(&self, node_id: NodeId, task_id: TaskId)
    -> DistResult<RecordBatchStream>;

    async fn get_job_status(
        &self,
        node_id: NodeId,
        job_id: Option<Uuid>,
    ) -> DistResult<HashMap<StageId, StageInfo>>;

    async fn cleanup_job(&self, node_id: NodeId, job_id: Uuid) -> DistResult<()>;
}

pub struct ScheduledTasks {
    pub stage_plans: HashMap<StageId, Arc<dyn ExecutionPlan>>,
    pub task_ids: Vec<TaskId>,
}

impl ScheduledTasks {
    pub fn new(
        stage_plans: HashMap<StageId, Arc<dyn ExecutionPlan>>,
        task_ids: Vec<TaskId>,
    ) -> Self {
        ScheduledTasks {
            stage_plans,
            task_ids,
        }
    }
}

#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct StageInfo {
    pub assigned_partitions: HashSet<usize>,
    pub task_set_infos: Vec<TaskSetInfo>,
}

impl StageInfo {
    pub async fn from_stage_state(stage_state: &StageState) -> Self {
        let task_set_infos = stage_state
            .task_sets
            .lock()
            .await
            .iter()
            .map(TaskSetInfo::from_task_set)
            .collect();

        StageInfo {
            assigned_partitions: stage_state.assigned_partitions.clone(),
            task_set_infos,
        }
    }
}

#[derive(Debug, Clone, Serialize, Deserialize)]
pub struct TaskSetInfo {
    pub running_partitions: HashSet<usize>,
    pub dropped_partitions: HashMap<usize, TaskMetrics>,
}

impl TaskSetInfo {
    pub fn from_task_set(task_set: &TaskSet) -> Self {
        TaskSetInfo {
            running_partitions: task_set.running_partitions.clone(),
            dropped_partitions: task_set.dropped_partitions.clone(),
        }
    }
}