sui_aws_orchestrator/
faults.rs

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
// Copyright (c) Mysten Labs, Inc.
// SPDX-License-Identifier: Apache-2.0

use std::{
    fmt::{Debug, Display},
    time::Duration,
};

use serde::{Deserialize, Serialize};

use crate::client::Instance;

#[derive(Clone, Serialize, Deserialize, Hash, PartialEq, Eq)]
pub enum FaultsType {
    /// Permanently crash the maximum number of nodes from the beginning.
    Permanent { faults: usize },
    /// Progressively crash and recover nodes.
    CrashRecovery {
        max_faults: usize,
        interval: Duration,
    },
}

impl Default for FaultsType {
    fn default() -> Self {
        Self::Permanent { faults: 0 }
    }
}

impl Debug for FaultsType {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Self::Permanent { faults } => write!(f, "{faults}"),
            Self::CrashRecovery {
                max_faults,
                interval,
            } => write!(f, "{max_faults}-{}cr", interval.as_secs()),
        }
    }
}

impl Display for FaultsType {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        match self {
            Self::Permanent { faults } => {
                if *faults == 0 {
                    write!(f, "no faults")
                } else {
                    write!(f, "{faults} crashed")
                }
            }
            Self::CrashRecovery {
                max_faults,
                interval,
            } => write!(f, "{max_faults} crash-recovery, {}s", interval.as_secs()),
        }
    }
}

/// The actions to apply to the testbed, i.e., which instances to crash and recover.
#[derive(Default)]
pub struct CrashRecoveryAction {
    /// The instances to boot.
    pub boot: Vec<Instance>,
    /// The instances to kill.
    pub kill: Vec<Instance>,
}

impl Display for CrashRecoveryAction {
    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
        let booted = self.boot.len();
        let killed = self.kill.len();

        if self.boot.is_empty() {
            write!(f, "{killed} node(s) killed")
        } else if self.kill.is_empty() {
            write!(f, "{booted} node(s) recovered")
        } else {
            write!(f, "{killed} node(s) killed and {booted} node(s) recovered")
        }
    }
}

impl CrashRecoveryAction {
    pub fn boot(instances: Vec<Instance>) -> Self {
        Self {
            boot: instances,
            kill: Vec::new(),
        }
    }

    pub fn kill(instances: Vec<Instance>) -> Self {
        Self {
            boot: Vec::new(),
            kill: instances,
        }
    }

    pub fn no_op() -> Self {
        Self::default()
    }
}

pub struct CrashRecoverySchedule {
    /// The number of faulty nodes and the crash-recovery pattern to follow.
    faults_type: FaultsType,
    /// The available instances.
    instances: Vec<Instance>,
    /// The current number of dead nodes.
    dead: usize,
}

impl CrashRecoverySchedule {
    pub fn new(faults_type: FaultsType, instances: Vec<Instance>) -> Self {
        Self {
            faults_type,
            instances,
            dead: 0,
        }
    }
    pub fn update(&mut self) -> CrashRecoveryAction {
        match &self.faults_type {
            // Permanently crash the specified number of nodes.
            FaultsType::Permanent { faults } => {
                if self.dead == 0 {
                    self.dead = *faults;
                    CrashRecoveryAction::kill(self.instances.clone().drain(0..*faults).collect())
                } else {
                    CrashRecoveryAction::no_op()
                }
            }

            // Periodically crash and recover nodes.
            FaultsType::CrashRecovery { max_faults, .. } => {
                let min_faults = max_faults / 3;

                // Recover all nodes if we already crashed them all.
                if self.dead == *max_faults {
                    let instances: Vec<_> = self.instances.clone().drain(0..*max_faults).collect();
                    self.dead = 0;
                    CrashRecoveryAction::boot(instances)
                }
                // Otherwise crash a few nodes at the time.
                else {
                    let (l, h) = if self.dead == 0 && min_faults != 0 {
                        (0, min_faults)
                    } else if self.dead == min_faults && min_faults != 0 {
                        (min_faults, 2 * min_faults)
                    } else {
                        (2 * min_faults, *max_faults)
                    };

                    let instances: Vec<_> = self.instances.clone().drain(l..h).collect();
                    self.dead += h - l;
                    CrashRecoveryAction::kill(instances)
                }
            }
        }
    }
}

#[cfg(test)]
mod faults_tests {
    use std::time::Duration;

    use crate::client::Instance;

    use super::{CrashRecoverySchedule, FaultsType};

    #[test]
    fn crash_recovery_1_fault() {
        let max_faults = 1;
        let interval = Duration::from_secs(60);
        let faulty = (0..max_faults)
            .map(|i| Instance::new_for_test(i.to_string()))
            .collect();
        let mut schedule = CrashRecoverySchedule::new(
            FaultsType::CrashRecovery {
                max_faults,
                interval,
            },
            faulty,
        );

        let action = schedule.update();
        assert_eq!(action.boot.len(), 0);
        assert_eq!(action.kill.len(), 1);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 1);
        assert_eq!(action.kill.len(), 0);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 0);
        assert_eq!(action.kill.len(), 1);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 1);
        assert_eq!(action.kill.len(), 0);
    }

    #[test]
    fn crash_recovery_2_faults() {
        let max_faults = 2;
        let interval = Duration::from_secs(60);
        let faulty = (0..max_faults)
            .map(|i| Instance::new_for_test(i.to_string()))
            .collect();
        let mut schedule = CrashRecoverySchedule::new(
            FaultsType::CrashRecovery {
                max_faults,
                interval,
            },
            faulty,
        );

        let action = schedule.update();
        assert_eq!(action.boot.len(), 0);
        assert_eq!(action.kill.len(), 2);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 2);
        assert_eq!(action.kill.len(), 0);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 0);
        assert_eq!(action.kill.len(), 2);

        let action = schedule.update();
        assert_eq!(action.boot.len(), 2);
        assert_eq!(action.kill.len(), 0);
    }

    #[test]
    fn crash_recovery() {
        let interval = Duration::from_secs(60);
        for i in 3..33 {
            let max_faults = i;
            let min_faults = max_faults / 3;

            let instances = (0..max_faults)
                .map(|i| Instance::new_for_test(i.to_string()))
                .collect();
            let mut schedule = CrashRecoverySchedule::new(
                FaultsType::CrashRecovery {
                    max_faults,
                    interval,
                },
                instances,
            );

            let action = schedule.update();
            assert_eq!(action.boot.len(), 0);
            assert_eq!(action.kill.len(), min_faults);

            let action = schedule.update();
            assert_eq!(action.boot.len(), 0);
            assert_eq!(action.kill.len(), min_faults);

            let action = schedule.update();
            assert_eq!(action.boot.len(), 0);
            assert_eq!(action.kill.len(), max_faults - 2 * min_faults);

            let action = schedule.update();
            assert_eq!(action.boot.len(), max_faults);
            assert_eq!(action.kill.len(), 0);

            let action = schedule.update();
            assert_eq!(action.boot.len(), 0);
            assert_eq!(action.kill.len(), min_faults);
        }
    }
}