orca-control 0.2.6-rc.2

Control plane: API server, reconciler, and cluster state management
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

use std::collections::HashMap;

use orca_core::types::RuntimeKind;

use super::*;

fn make_node(id: u64, cpu: f64, mem: u64, workloads: u32) -> NodeCapacity {
    NodeCapacity {
        node_id: id,
        cpu_available: cpu,
        memory_available: mem,
        gpu_count: 0,
        gpu_vram_available: 0,
        has_wasm_runtime: false,
        labels: HashMap::new(),
        current_workload_count: workloads,
    }
}

fn make_service(name: &str, replicas: u32) -> ServiceRequest {
    ServiceRequest {
        name: name.to_string(),
        replicas_desired: replicas,
        runtime: RuntimeKind::Container,
        cpu_required: 1.0,
        memory_required: 512,
        gpu_required: 0,
        placement_labels: HashMap::new(),
        requires_gpu: false,
    }
}

#[test]
fn spreads_replicas_across_nodes() {
    let nodes = vec![
        make_node(1, 4.0, 4096, 0),
        make_node(2, 4.0, 4096, 0),
        make_node(3, 4.0, 4096, 0),
    ];
    let svc = make_service("web", 3);
    let actions = schedule(&[svc], &nodes, &[]);

    let mut assigned_nodes: Vec<u64> = actions
        .iter()
        .filter_map(|a| match a {
            ScheduleAction::Assign { node_id, .. } => Some(*node_id),
            _ => None,
        })
        .collect();
    assigned_nodes.sort();
    assert_eq!(assigned_nodes, vec![1, 2, 3]);
}

#[test]
fn gpu_requirement_filters_nodes() {
    let mut gpu_node = make_node(1, 4.0, 4096, 0);
    gpu_node.gpu_count = 2;
    gpu_node.gpu_vram_available = 16_000;

    let cpu_node = make_node(2, 8.0, 8192, 0);

    let svc = ServiceRequest {
        name: "ml-train".to_string(),
        replicas_desired: 1,
        runtime: RuntimeKind::Container,
        cpu_required: 1.0,
        memory_required: 512,
        gpu_required: 1,
        placement_labels: HashMap::new(),
        requires_gpu: true,
    };

    let actions = schedule(&[svc], &[gpu_node, cpu_node], &[]);
    assert_eq!(actions.len(), 1);
    match &actions[0] {
        ScheduleAction::Assign { node_id, .. } => assert_eq!(*node_id, 1),
        _ => panic!("expected Assign action"),
    }
}

#[test]
fn scale_down_removes_from_most_loaded() {
    let nodes = vec![make_node(1, 4.0, 4096, 5), make_node(2, 4.0, 4096, 1)];
    let svc = make_service("api", 1);
    let assignments = vec![
        Assignment {
            service: "api".to_string(),
            replica_idx: 0,
            node_id: 1,
        },
        Assignment {
            service: "api".to_string(),
            replica_idx: 1,
            node_id: 2,
        },
    ];

    let actions = schedule(&[svc], &nodes, &assignments);
    assert_eq!(actions.len(), 1);
    match &actions[0] {
        ScheduleAction::Unassign { node_id, .. } => assert_eq!(*node_id, 1),
        _ => panic!("expected Unassign action"),
    }
}

#[test]
fn wasm_preference_scores_higher() {
    let mut wasm_node = make_node(1, 4.0, 4096, 0);
    wasm_node.has_wasm_runtime = true;

    let container_node = make_node(2, 4.0, 4096, 0);

    let svc = ServiceRequest {
        name: "edge-fn".to_string(),
        replicas_desired: 1,
        runtime: RuntimeKind::Wasm,
        cpu_required: 0.5,
        memory_required: 256,
        gpu_required: 0,
        placement_labels: HashMap::new(),
        requires_gpu: false,
    };

    let actions = schedule(&[svc], &[wasm_node, container_node], &[]);
    assert_eq!(actions.len(), 1);
    match &actions[0] {
        ScheduleAction::Assign { node_id, .. } => assert_eq!(*node_id, 1),
        _ => panic!("expected Assign action"),
    }
}