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
use crate::transfer_status::TransferStatus;
use crate::inner::scheduler_state::SchedulerState;
use crate::inner::worker_event::WorkerEvent;
pub(crate) async fn handle_worker_event(event: WorkerEvent, state: &mut SchedulerState) {
match event {
WorkerEvent::Progress {
key,
next_offset,
total_size,
} => {
crate::meow_trace_log!(
"worker_event",
"progress: key={} next_offset={} total_size={}",
crate::inner::safe_key(&key),
next_offset,
total_size
);
// Persist/emit progress only for a still-live group. Gating the
// insert on the live group prevents a late or out-of-order Progress
// (possible once parts run concurrently) from re-creating an offsets
// entry that a terminal event already removed. The single-sender
// discipline in run_group already orders events, so this is
// defense-in-depth and is a no-op for the serial path (where Progress
// always precedes the terminal event).
if state.groups().contains_key(&key) {
state.offsets_mut().insert(key.clone(), next_offset);
if let Some(group) = state.groups().get(&key) {
crate::inner::exec_impl::emit::emit_status(
state,
group.entry(),
TransferStatus::Transmission,
next_offset,
total_size,
);
}
} else {
crate::log::emit_lazy(|| {
crate::log::Log::warn(
"worker_event",
format!(
"stray Progress for dead group; ignored next_offset={} total_size={}",
next_offset, total_size
),
)
.with_key(key.1.as_str())
.with_offset(next_offset)
});
}
}
WorkerEvent::Completed {
key,
total_size,
completion_payload,
} => {
crate::meow_key_log!(
"worker_event",
"completed: key={} total_size={}",
crate::inner::safe_key(&key),
total_size
);
state.active_mut().remove(&key);
// 完成后无论此前是否 paused，都要清理 paused 标记。
state.paused_set_mut().remove(&key);
state.offsets_mut().insert(key.clone(), total_size);
if let Some(group) = state.groups_mut().remove(&key) {
state
.task_id_to_dedupe_mut()
.remove(&group.leader_inner().task_id());
let task_id = group.entry().inner().task_id();
crate::inner::exec_impl::emit::emit_status(
state,
group.entry(),
TransferStatus::Complete,
total_size,
total_size,
);
if let Some(cb) = group.entry().callbacks().complete_cb() {
crate::inner::exec_impl::emit::invoke_complete_cb(
state,
cb,
task_id,
completion_payload,
);
}
} else {
crate::log::emit_lazy(|| {
crate::log::Log::warn(
"worker_event",
format!("Completed for unknown group; nothing to finalize total_size={}", total_size),
)
.with_key(key.1.as_str())
});
}
state.offsets_mut().remove(&key);
}
WorkerEvent::Failed { key, error } => {
state.active_mut().remove(&key);
// 失败终态会结束任务生命周期，因此同步清理 paused 标记。
state.paused_set_mut().remove(&key);
if let Some(group) = state.groups_mut().remove(&key) {
state
.task_id_to_dedupe_mut()
.remove(&group.leader_inner().task_id());
let current = state.offsets().get(&key).copied().unwrap_or(0);
// Core chunk/part-failure log: terminal task failure with the
// contiguous offset reached and the underlying error detail.
// Emitted before `error` is moved into `TransferStatus::Failed`.
crate::log::emit_lazy(|| {
let mut log = crate::log::Log::error(
"worker_event",
format!("task failed: {}", crate::log::redact_secrets(&error.to_string())),
)
.with_key(key.1.as_str())
.with_offset(current)
.with_error_code(error.code());
if let Some(status) = error.http_status() {
log = log.with_http_status(status);
}
log
});
crate::inner::exec_impl::emit::emit_status(
state,
group.entry(),
TransferStatus::Failed(error),
current,
group.entry().inner().total_size(),
);
} else {
crate::log::emit_lazy(|| {
crate::log::Log::warn(
"worker_event",
format!("Failed for unknown group; nothing to finalize error={}", crate::log::redact_secrets(&error.to_string())),
)
.with_key(key.1.as_str())
.with_error_code(error.code())
});
}
state.offsets_mut().remove(&key);
}
WorkerEvent::Canceled { key } => {
crate::meow_key_log!("worker_event", "canceled: key={}", crate::inner::safe_key(&key));
state.active_mut().remove(&key);
// 若 key 在 paused_set 中，表示该取消来自 pause 流程，仅收敛执行态，不销毁 group。
if state.paused_set().contains(&key) {
crate::meow_key_log!(
"worker_event",
"canceled from pause flow; keep group for resume: key={}",
crate::inner::safe_key(&key)
);
return;
}
if let Some(group) = state.groups_mut().remove(&key) {
state
.task_id_to_dedupe_mut()
.remove(&group.leader_inner().task_id());
let current = state.offsets().get(&key).copied().unwrap_or(0);
crate::inner::exec_impl::emit::emit_status(
state,
group.entry(),
TransferStatus::Canceled,
current,
group.entry().inner().total_size(),
);
} else {
crate::log::emit_lazy(|| {
crate::log::Log::warn(
"worker_event",
"Canceled for unknown group; nothing to finalize".to_string(),
)
.with_key(key.1.as_str())
});
}
state.offsets_mut().remove(&key);
}
}
}
#[cfg(test)]
mod tests {
use std::sync::{Arc, RwLock};
use super::handle_worker_event;
use crate::dflt::default_http_transfer::default_breakpoint_arcs;
use crate::http_breakpoint::BreakpointDownloadHttpConfig;
use crate::inner::cb_dispatcher;
use crate::inner::group_state::{GroupState, RecordEntry};
use crate::inner::inner_task::InnerTask;
use crate::inner::scheduler_state::SchedulerState;
use crate::inner::task_callbacks::TaskCallbacks;
use crate::inner::worker_event::WorkerEvent;
use crate::inner::UniqueId;
use crate::up_pounce_builder::UploadPounceBuilder;
/// Builds a SchedulerState holding one live upload group, exactly as the
/// scheduler would after the task started (group + offsets + task_id map
/// populated). Returns the state, the group key, its total size, and the
/// dispatcher join guard (kept alive so callback submission stays connected).
async fn live_upload_state() -> (SchedulerState, UniqueId, u64) {
let mut path = std::env::temp_dir();
let ts = std::time::SystemTime::now()
.duration_since(std::time::UNIX_EPOCH)
.expect("clock")
.as_nanos();
path.push(format!("rusty_cat_offsets_teardown_{ts}.bin"));
std::fs::write(&path, vec![7u8; 4096]).expect("write fixture");
let pounce = UploadPounceBuilder::new("teardown.bin", &path, 1024)
.with_url("https://placeholder/upload")
.build()
.expect("build pounce");
let (def_up, def_down) = default_breakpoint_arcs();
let inner = InnerTask::from_pounce(
pounce,
BreakpointDownloadHttpConfig::default(),
None,
def_up,
def_down,
)
.await
.expect("from_pounce");
let _ = std::fs::remove_file(&path);
let key = inner.dedupe_key();
let total = inner.total_size();
let (cb_submit, cb_join) = cb_dispatcher::start().expect("start dispatcher");
// Detach the dispatcher join guard. Its `Drop` blocks on `join`, which
// only returns once the channel's sole sender (held inside `state`) is
// dropped. Relying on local drop order is fragile in a test that may
// panic mid-assertion (the guard would join while the sender is still
// alive — a deadlock). Forgetting it is safe: the dispatcher thread still
// exits on its own when `state` drops and closes the channel.
std::mem::forget(cb_join);
let mut state = SchedulerState::new(1, 1, Arc::new(RwLock::new(Vec::new())), cb_submit);
state
.task_id_to_dedupe_mut()
.insert(inner.task_id(), key.clone());
state.offsets_mut().insert(key.clone(), 0);
let entry = RecordEntry::new(inner.clone(), TaskCallbacks::empty());
state
.groups_mut()
.insert(key.clone(), GroupState::new(inner.clone(), entry));
(state, key, total)
}
/// MISSED-D: once a terminal event has torn down a group, a late or
/// out-of-order `Progress` (possible only once parts run concurrently) must
/// NOT re-create an orphan `offsets` entry. The insert is gated on the group
/// still being live.
#[tokio::test]
async fn stray_progress_after_complete_creates_no_orphan_offset() {
let (mut state, key, total) = live_upload_state().await;
assert!(state.groups().contains_key(&key));
// Terminal Completed tears the group down and clears its offsets entry.
handle_worker_event(
WorkerEvent::Completed {
key: key.clone(),
total_size: total,
completion_payload: None,
},
&mut state,
)
.await;
assert!(
!state.groups().contains_key(&key),
"group must be removed after Completed"
);
assert!(
state.offsets().get(&key).is_none(),
"offsets entry must be cleared after Completed"
);
// A stray Progress for the now-dead group must be a no-op for offsets.
handle_worker_event(
WorkerEvent::Progress {
key: key.clone(),
next_offset: 512,
total_size: total,
},
&mut state,
)
.await;
assert!(
state.offsets().get(&key).is_none(),
"stray Progress after teardown must not resurrect an orphan offsets entry"
);
}
/// A Progress for a STILL-LIVE group does persist (proves the guard is not
/// over-broad and the serial path is unaffected).
#[tokio::test]
async fn progress_for_live_group_persists_offset() {
let (mut state, key, total) = live_upload_state().await;
handle_worker_event(
WorkerEvent::Progress {
key: key.clone(),
next_offset: 1024,
total_size: total,
},
&mut state,
)
.await;
assert_eq!(
state.offsets().get(&key).copied(),
Some(1024),
"progress for a live group must persist its contiguous offset"
);
}
}