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
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
|
#include "../lib/cluster.h"
#include "../lib/runner.h"
struct fixture
{
FIXTURE_CLUSTER;
};
static void *setUp(const MunitParameter params[], MUNIT_UNUSED void *user_data)
{
struct fixture *f = munit_malloc(sizeof *f);
SETUP_CLUSTER();
return f;
}
static void tearDown(void *data)
{
struct fixture *f = data;
TEAR_DOWN_CLUSTER();
free(f);
}
/******************************************************************************
*
* raft_transfer
*
*****************************************************************************/
SUITE(raft_transfer)
/* The follower we ask to transfer leadership to is up-to-date. */
TEST(raft_transfer, UpToDate, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 2 voters. */
for (id = 1; id <= 2; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 2 /* servers */, 2 /* voters */);
CLUSTER_START(id);
}
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" probe server 2 sending a heartbeat (no entries)\n"
"[ 130] 2 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 140] 1 > recv append entries result from server 2\n");
test_cluster_transfer(&f->cluster_, 1, 2);
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 2);
CLUSTER_TRACE(
"[ 140] 1 > transfer leadership to 2\n"
" send timeout to 2\n"
"[ 150] 2 > recv timeout now from server 1\n"
" convert to candidate, start election for term 3\n"
"[ 160] 1 > recv request vote from server 2\n"
" remote term is higher (3 vs 2) -> bump term, step down\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 170] 2 > recv request vote result from server 1\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" probe server 1 sending a heartbeat (no entries)\n");
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 0);
return MUNIT_OK;
}
/* The follower we ask to transfer leadership to needs to catch up. */
TEST(raft_transfer, CatchUp, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 2 voters. Server 1 has an additioanl
* entry. */
for (id = 1; id <= 2; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 2 /* servers */, 2 /* voters */);
if (id == 1) {
CLUSTER_ADD_ENTRY(id, RAFT_COMMAND, 1 /* term */, 0 /* payload */);
}
CLUSTER_START(id);
}
/* Server 1 becomes leader. */
CLUSTER_TRACE(
"[ 0] 1 > term 1, 2 entries (1^1..2^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is longer (2^1 vs 1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" replicate 1 new barrier entry (3^2)\n"
" probe server 2 sending 1 entry (3^2)\n");
/* Fire a transfer event. Since server 2 is not up-to-date, server 1 will
* not send it a TimeoutNow message immediately. */
test_cluster_transfer(&f->cluster_, 1, 2);
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 2);
CLUSTER_TRACE(
"[ 120] 1 > transfer leadership to 2\n"
" wait for transferee to catch up\n"
"[ 130] 1 > persisted 1 entry (3^2)\n"
" next uncommitted entry (2^1) has 1 vote out of 2\n"
"[ 130] 2 > recv append entries from server 1\n"
" missing previous entry (2^1) -> reject\n"
"[ 140] 1 > recv append entries result from server 2\n"
" log mismatch -> send old entries\n"
" probe server 2 sending 2 entries (2^1..3^2)\n"
"[ 150] 2 > recv append entries from server 1\n"
" start persisting 2 new entries (2^1..3^2)\n"
"[ 160] 2 > persisted 2 entry (2^1..3^2)\n"
" send success result to 1\n");
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 2);
/* Server 2 is now up-to-date, so server 1 sends it a timeout. */
CLUSTER_TRACE(
"[ 170] 1 > recv append entries result from server 2\n"
" send timeout to 2\n"
" commit 2 new entries (2^1..3^2)\n"
"[ 180] 2 > recv timeout now from server 1\n"
" convert to candidate, start election for term 3\n"
"[ 190] 1 > recv request vote from server 2\n"
" remote term is higher (3 vs 2) -> bump term, step down\n"
" remote log is equal (3^2) -> grant vote\n"
"[ 200] 2 > recv request vote result from server 1\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" replicate 1 new barrier entry (4^3)\n"
" probe server 1 sending 1 entry (4^3)\n");
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 0);
return MUNIT_OK;
}
/* The follower we ask to transfer leadership to is down and the leadership
* transfer does not succeed. */
TEST(raft_transfer, Expire, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 3 voters. */
for (id = 1; id <= 3; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 3 /* servers */, 3 /* voters */);
CLUSTER_START(id);
}
/* Server 1 becomes leader. */
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 0] 3 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 110] 3 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 3 -> convert to leader\n"
" probe server 2 sending a heartbeat (no entries)\n"
" probe server 3 sending a heartbeat (no entries)\n"
"[ 120] 1 > recv request vote result from server 3\n"
" local server is leader -> ignore\n"
"[ 130] 2 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 130] 3 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 140] 1 > recv append entries result from server 2\n"
"[ 140] 1 > recv append entries result from server 3\n");
/* Stop server 2 and try to transfer leadership to it. */
CLUSTER_STOP(2 /* ID */);
test_cluster_transfer(&f->cluster_, 1, 2);
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 2);
/* Eventually server 1 stops trying to transfer its leadership. */
CLUSTER_TRACE(
"[ 140] 1 > transfer leadership to 2\n"
" send timeout to 2\n"
"[ 170] 1 > timeout as leader\n"
" pipeline server 2 sending a heartbeat (no entries)\n"
" pipeline server 3 sending a heartbeat (no entries)\n"
"[ 180] 3 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 190] 1 > recv append entries result from server 3\n"
"[ 220] 1 > timeout as leader\n"
" server 2 is unreachable -> abort pipeline\n"
" probe server 2 sending a heartbeat (no entries)\n"
" pipeline server 3 sending a heartbeat (no entries)\n"
"[ 230] 3 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 240] 1 > recv append entries result from server 3\n"
"[ 270] 1 > timeout as leader\n"
" probe server 2 sending a heartbeat (no entries)\n"
" pipeline server 3 sending a heartbeat (no entries)\n"
" server 2 not replicating fast enough -> abort transfer\n");
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 0);
return MUNIT_OK;
}
/* The given ID doesn't match any server in the current configuration. */
TEST(raft_transfer, UnknownServer, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
struct raft_event event;
struct raft_update update;
int rv;
/* Start a single-node cluster. */
CLUSTER_SET_TERM(1 /* ID */, 1 /* term */);
CLUSTER_ADD_ENTRY(1 /* ID */, RAFT_CHANGE, 1 /* servers */, 1 /* voters */);
CLUSTER_START(1 /* ID */);
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
" self elect and convert to leader\n");
event.time = f->cluster_.time;
event.type = RAFT_TRANSFER;
event.transfer.server_id = 2;
rv = raft_step(CLUSTER_RAFT(1), &event, &update);
munit_assert_int(rv, ==, RAFT_BADID);
return MUNIT_OK;
}
/* Submitting a transfer request twice is an error. */
TEST(raft_transfer, Twice, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
struct raft_event event;
struct raft_update update;
unsigned id;
int rv;
/* Bootstrap and start a cluster with 2 voters. */
for (id = 1; id <= 2; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 2 /* servers */, 2 /* voters */);
CLUSTER_START(id);
}
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" probe server 2 sending a heartbeat (no entries)\n");
test_cluster_transfer(&f->cluster_, 1, 2);
event.time = f->cluster_.time;
event.type = RAFT_TRANSFER;
event.transfer.server_id = 2;
rv = raft_step(CLUSTER_RAFT(1), &event, &update);
munit_assert_int(rv, ==, RAFT_NOTLEADER);
return MUNIT_OK;
}
/* If the given ID is zero, the target is selected automatically. */
TEST(raft_transfer, AutoSelect, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 3 voters. */
for (id = 1; id <= 3; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 3 /* servers */, 3 /* voters */);
CLUSTER_START(id);
}
/* Server 1 becomes leader. */
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 0] 3 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 110] 3 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 3 -> convert to leader\n"
" probe server 2 sending a heartbeat (no entries)\n"
" probe server 3 sending a heartbeat (no entries)\n"
"[ 120] 1 > recv request vote result from server 3\n"
" local server is leader -> ignore\n"
"[ 130] 2 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 130] 3 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 140] 1 > recv append entries result from server 2\n");
test_cluster_transfer(&f->cluster_, 1, 0);
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 2);
return MUNIT_OK;
}
/* If the given ID is zero, the target is selected automatically. Followers that
* are up-to-date are preferred. */
TEST(raft_transfer, AutoSelectUpToDate, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 3 voters. Server 1 has an additional
* entry. */
for (id = 1; id <= 3; id++) {
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 3 /* servers */, 3 /* voters */);
if (id == 1) {
CLUSTER_ADD_ENTRY(id, RAFT_COMMAND, 1 /* term */, 0 /* payload */);
}
CLUSTER_START(id);
}
/* Server 1 becomes leader and starts replicating the additional entry. */
CLUSTER_TRACE(
"[ 0] 1 > term 1, 2 entries (1^1..2^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 0] 3 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is longer (2^1 vs 1^1) -> grant vote\n"
"[ 110] 3 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is longer (2^1 vs 1^1) -> grant vote\n"
"[ 120] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 3 -> convert to leader\n"
" replicate 1 new barrier entry (3^2)\n"
" probe server 2 sending 1 entry (3^2)\n"
" probe server 3 sending 1 entry (3^2)\n");
/* Stop server 2, so it won't get the extra entry, while server 3 does. */
CLUSTER_STOP(2 /* ID */);
CLUSTER_TRACE(
"[ 120] 1 > recv request vote result from server 3\n"
" local server is leader -> ignore\n"
"[ 130] 1 > persisted 1 entry (3^2)\n"
" next uncommitted entry (2^1) has 1 vote out of 3\n"
"[ 130] 3 > recv append entries from server 1\n"
" missing previous entry (2^1) -> reject\n"
"[ 140] 1 > recv append entries result from server 3\n"
" log mismatch -> send old entries\n"
" probe server 3 sending 2 entries (2^1..3^2)\n"
"[ 150] 3 > recv append entries from server 1\n"
" start persisting 2 new entries (2^1..3^2)\n"
"[ 160] 3 > persisted 2 entry (2^1..3^2)\n"
" send success result to 1\n"
"[ 170] 1 > recv append entries result from server 3\n"
" commit 2 new entries (2^1..3^2)\n");
/* The auto-selection logic prefers server 3, since it's more up-to-date. */
test_cluster_transfer(&f->cluster_, 1, 0);
munit_assert_ullong(raft_transferee(CLUSTER_RAFT(1)), ==, 3);
return MUNIT_OK;
}
/* It's possible to transfer leadership also when pre-vote is active */
TEST(raft_transfer, PreVote, setUp, tearDown, 0, NULL)
{
struct fixture *f = data;
unsigned id;
/* Bootstrap and start a cluster with 2 voters, enabling pre-vote.*/
for (id = 1; id <= 2; id++) {
raft_set_pre_vote(CLUSTER_RAFT(id), true);
CLUSTER_SET_TERM(id, 1 /* term */);
CLUSTER_ADD_ENTRY(id, RAFT_CHANGE, 2 /* servers */, 2 /* voters */);
CLUSTER_START(id);
}
/* Server 1 becomes leader. */
CLUSTER_TRACE(
"[ 0] 1 > term 1, 1 entry (1^1)\n"
"[ 0] 2 > term 1, 1 entry (1^1)\n"
"[ 100] 1 > timeout as follower\n"
" convert to candidate, start pre-election for term 2\n"
"[ 110] 2 > recv request vote from server 1\n"
" remote log is equal (1^1) -> pre-vote ok\n"
"[ 120] 1 > recv request vote result from server 2\n"
" votes quorum reached -> pre-vote successful\n"
"[ 130] 2 > recv request vote from server 1\n"
" remote term is higher (2 vs 1) -> bump term\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 140] 1 > recv request vote result from server 2\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" probe server 2 sending a heartbeat (no entries)\n"
"[ 150] 2 > recv append entries from server 1\n"
" no new entries to persist\n"
"[ 160] 1 > recv append entries result from server 2\n");
/* Perform a successful leadership transfer. */
test_cluster_transfer(&f->cluster_, 1, 2);
CLUSTER_TRACE(
"[ 160] 1 > transfer leadership to 2\n"
" send timeout to 2\n"
"[ 170] 2 > recv timeout now from server 1\n"
" convert to candidate, start election for term 3\n"
"[ 180] 1 > recv request vote from server 2\n"
" remote term is higher (3 vs 2) -> bump term, step down\n"
" remote log is equal (1^1) -> grant vote\n"
"[ 190] 2 > recv request vote result from server 1\n"
" quorum reached with 2 votes out of 2 -> convert to leader\n"
" probe server 1 sending a heartbeat (no entries)\n");
return MUNIT_OK;
}
|
