forked from redis/redis
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcluster.c
4498 lines (4007 loc) · 173 KB
/
cluster.c
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
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
/* Redis Cluster implementation.
*
* Copyright (c) 2009-2012, Salvatore Sanfilippo <antirez at gmail dot com>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* * Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
* * Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* * Neither the name of Redis nor the names of its contributors may be used
* to endorse or promote products derived from this software without
* specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
#include "redis.h"
#include "cluster.h"
#include "endianconv.h"
#include <sys/types.h>
#include <sys/socket.h>
#include <arpa/inet.h>
#include <fcntl.h>
#include <unistd.h>
#include <sys/socket.h>
#include <sys/stat.h>
#include <sys/file.h>
/* A global reference to myself is handy to make code more clear.
* Myself always points to server.cluster->myself, that is, the clusterNode
* that represents this node. */
clusterNode *myself = NULL;
clusterNode *createClusterNode(char *nodename, int flags);
int clusterAddNode(clusterNode *node);
void clusterAcceptHandler(aeEventLoop *el, int fd, void *privdata, int mask);
void clusterReadHandler(aeEventLoop *el, int fd, void *privdata, int mask);
void clusterSendPing(clusterLink *link, int type);
void clusterSendFail(char *nodename);
void clusterSendFailoverAuthIfNeeded(clusterNode *node, clusterMsg *request);
void clusterUpdateState(void);
int clusterNodeGetSlotBit(clusterNode *n, int slot);
sds clusterGenNodesDescription(int filter);
clusterNode *clusterLookupNode(char *name);
int clusterNodeAddSlave(clusterNode *master, clusterNode *slave);
int clusterAddSlot(clusterNode *n, int slot);
int clusterDelSlot(int slot);
int clusterDelNodeSlots(clusterNode *node);
int clusterNodeSetSlotBit(clusterNode *n, int slot);
void clusterSetMaster(clusterNode *n);
void clusterHandleSlaveFailover(void);
void clusterHandleSlaveMigration(int max_slaves);
int bitmapTestBit(unsigned char *bitmap, int pos);
void clusterDoBeforeSleep(int flags);
void clusterSendUpdate(clusterLink *link, clusterNode *node);
void resetManualFailover(void);
void clusterCloseAllSlots(void);
void clusterSetNodeAsMaster(clusterNode *n);
void clusterDelNode(clusterNode *delnode);
/* -----------------------------------------------------------------------------
* Initialization
* -------------------------------------------------------------------------- */
/* Return the greatest configEpoch found in the cluster. */
uint64_t clusterGetMaxEpoch(void) {
uint64_t max = 0;
dictIterator *di;
dictEntry *de;
di = dictGetSafeIterator(server.cluster->nodes);
while((de = dictNext(di)) != NULL) {
clusterNode *node = dictGetVal(de);
if (node->configEpoch > max) max = node->configEpoch;
}
dictReleaseIterator(di);
if (max < server.cluster->currentEpoch) max = server.cluster->currentEpoch;
return max;
}
/* Load the cluster config from 'filename'.
*
* If the file does not exist or is zero-length (this may happen because
* when we lock the nodes.conf file, we create a zero-length one for the
* sake of locking if it does not already exist), REDIS_ERR is returned.
* If the configuration was loaded from the file, REDIS_OK is returned. */
int clusterLoadConfig(char *filename) {
FILE *fp = fopen(filename,"r");
struct stat sb;
char *line;
int maxline, j;
if (fp == NULL) {
if (errno == ENOENT) {
return REDIS_ERR;
} else {
redisLog(REDIS_WARNING,
"Loading the cluster node config from %s: %s",
filename, strerror(errno));
exit(1);
}
}
/* Check if the file is zero-length: if so return REDIS_ERR to signal
* we have to write the config. */
if (fstat(fileno(fp),&sb) != -1 && sb.st_size == 0) {
fclose(fp);
return REDIS_ERR;
}
/* Parse the file. Note that single liens of the cluster config file can
* be really long as they include all the hash slots of the node.
* This means in the worst possible case, half of the Redis slots will be
* present in a single line, possibly in importing or migrating state, so
* together with the node ID of the sender/receiver.
*
* To simplify we allocate 1024+REDIS_CLUSTER_SLOTS*128 bytes per line. */
maxline = 1024+REDIS_CLUSTER_SLOTS*128;
line = zmalloc(maxline);
while(fgets(line,maxline,fp) != NULL) {
int argc;
sds *argv;
clusterNode *n, *master;
char *p, *s;
/* Skip blank lines, they can be created either by users manually
* editing nodes.conf or by the config writing process if stopped
* before the truncate() call. */
if (line[0] == '\n') continue;
/* Split the line into arguments for processing. */
argv = sdssplitargs(line,&argc);
if (argv == NULL) goto fmterr;
/* Handle the special "vars" line. Don't pretend it is the last
* line even if it actually is when generated by Redis. */
if (strcasecmp(argv[0],"vars") == 0) {
for (j = 1; j < argc; j += 2) {
if (strcasecmp(argv[j],"currentEpoch") == 0) {
server.cluster->currentEpoch =
strtoull(argv[j+1],NULL,10);
} else if (strcasecmp(argv[j],"lastVoteEpoch") == 0) {
server.cluster->lastVoteEpoch =
strtoull(argv[j+1],NULL,10);
} else {
redisLog(REDIS_WARNING,
"Skipping unknown cluster config variable '%s'",
argv[j]);
}
}
continue;
}
/* Create this node if it does not exist */
n = clusterLookupNode(argv[0]);
if (!n) {
n = createClusterNode(argv[0],0);
clusterAddNode(n);
}
/* Address and port */
if ((p = strchr(argv[1],':')) == NULL) goto fmterr;
*p = '\0';
memcpy(n->ip,argv[1],strlen(argv[1])+1);
n->port = atoi(p+1);
/* Parse flags */
p = s = argv[2];
while(p) {
p = strchr(s,',');
if (p) *p = '\0';
if (!strcasecmp(s,"myself")) {
redisAssert(server.cluster->myself == NULL);
myself = server.cluster->myself = n;
n->flags |= REDIS_NODE_MYSELF;
} else if (!strcasecmp(s,"master")) {
n->flags |= REDIS_NODE_MASTER;
} else if (!strcasecmp(s,"slave")) {
n->flags |= REDIS_NODE_SLAVE;
} else if (!strcasecmp(s,"fail?")) {
n->flags |= REDIS_NODE_PFAIL;
} else if (!strcasecmp(s,"fail")) {
n->flags |= REDIS_NODE_FAIL;
n->fail_time = mstime();
} else if (!strcasecmp(s,"handshake")) {
n->flags |= REDIS_NODE_HANDSHAKE;
} else if (!strcasecmp(s,"noaddr")) {
n->flags |= REDIS_NODE_NOADDR;
} else if (!strcasecmp(s,"noflags")) {
/* nothing to do */
} else {
redisPanic("Unknown flag in redis cluster config file");
}
if (p) s = p+1;
}
/* Get master if any. Set the master and populate master's
* slave list. */
if (argv[3][0] != '-') {
master = clusterLookupNode(argv[3]);
if (!master) {
master = createClusterNode(argv[3],0);
clusterAddNode(master);
}
n->slaveof = master;
clusterNodeAddSlave(master,n);
}
/* Set ping sent / pong received timestamps */
if (atoi(argv[4])) n->ping_sent = mstime();
if (atoi(argv[5])) n->pong_received = mstime();
/* Set configEpoch for this node. */
n->configEpoch = strtoull(argv[6],NULL,10);
/* Populate hash slots served by this instance. */
for (j = 8; j < argc; j++) {
int start, stop;
if (argv[j][0] == '[') {
/* Here we handle migrating / importing slots */
int slot;
char direction;
clusterNode *cn;
p = strchr(argv[j],'-');
redisAssert(p != NULL);
*p = '\0';
direction = p[1]; /* Either '>' or '<' */
slot = atoi(argv[j]+1);
p += 3;
cn = clusterLookupNode(p);
if (!cn) {
cn = createClusterNode(p,0);
clusterAddNode(cn);
}
if (direction == '>') {
server.cluster->migrating_slots_to[slot] = cn;
} else {
server.cluster->importing_slots_from[slot] = cn;
}
continue;
} else if ((p = strchr(argv[j],'-')) != NULL) {
*p = '\0';
start = atoi(argv[j]);
stop = atoi(p+1);
} else {
start = stop = atoi(argv[j]);
}
while(start <= stop) clusterAddSlot(n, start++);
}
sdsfreesplitres(argv,argc);
}
zfree(line);
fclose(fp);
/* Config sanity check */
redisAssert(server.cluster->myself != NULL);
redisLog(REDIS_NOTICE,"Node configuration loaded, I'm %.40s", myself->name);
/* Something that should never happen: currentEpoch smaller than
* the max epoch found in the nodes configuration. However we handle this
* as some form of protection against manual editing of critical files. */
if (clusterGetMaxEpoch() > server.cluster->currentEpoch) {
server.cluster->currentEpoch = clusterGetMaxEpoch();
}
return REDIS_OK;
fmterr:
redisLog(REDIS_WARNING,
"Unrecoverable error: corrupted cluster config file.");
fclose(fp);
exit(1);
}
/* Cluster node configuration is exactly the same as CLUSTER NODES output.
*
* This function writes the node config and returns 0, on error -1
* is returned.
*
* Note: we need to write the file in an atomic way from the point of view
* of the POSIX filesystem semantics, so that if the server is stopped
* or crashes during the write, we'll end with either the old file or the
* new one. Since we have the full payload to write available we can use
* a single write to write the whole file. If the pre-existing file was
* bigger we pad our payload with newlines that are anyway ignored and truncate
* the file afterward. */
int clusterSaveConfig(int do_fsync) {
sds ci;
size_t content_size;
struct stat sb;
int fd;
server.cluster->todo_before_sleep &= ~CLUSTER_TODO_SAVE_CONFIG;
/* Get the nodes description and concatenate our "vars" directive to
* save currentEpoch and lastVoteEpoch. */
ci = clusterGenNodesDescription(REDIS_NODE_HANDSHAKE);
ci = sdscatprintf(ci,"vars currentEpoch %llu lastVoteEpoch %llu\n",
(unsigned long long) server.cluster->currentEpoch,
(unsigned long long) server.cluster->lastVoteEpoch);
content_size = sdslen(ci);
if ((fd = open(server.cluster_configfile,O_WRONLY|O_CREAT,0644))
== -1) goto err;
/* Pad the new payload if the existing file length is greater. */
if (fstat(fd,&sb) != -1) {
if (sb.st_size > content_size) {
ci = sdsgrowzero(ci,sb.st_size);
memset(ci+content_size,'\n',sb.st_size-content_size);
}
}
if (write(fd,ci,sdslen(ci)) != (ssize_t)sdslen(ci)) goto err;
if (do_fsync) {
server.cluster->todo_before_sleep &= ~CLUSTER_TODO_FSYNC_CONFIG;
fsync(fd);
}
/* Truncate the file if needed to remove the final \n padding that
* is just garbage. */
if (content_size != sdslen(ci) && ftruncate(fd,content_size) == -1) {
/* ftruncate() failing is not a critical error. */
}
close(fd);
sdsfree(ci);
return 0;
err:
if (fd != -1) close(fd);
sdsfree(ci);
return -1;
}
void clusterSaveConfigOrDie(int do_fsync) {
if (clusterSaveConfig(do_fsync) == -1) {
redisLog(REDIS_WARNING,"Fatal: can't update cluster config file.");
exit(1);
}
}
/* Lock the cluster config using flock(), and leaks the file descritor used to
* acquire the lock so that the file will be locked forever.
*
* This works because we always update nodes.conf with a new version
* in-place, reopening the file, and writing to it in place (later adjusting
* the length with ftruncate()).
*
* On success REDIS_OK is returned, otherwise an error is logged and
* the function returns REDIS_ERR to signal a lock was not acquired. */
int clusterLockConfig(char *filename) {
/* To lock it, we need to open the file in a way it is created if
* it does not exist, otherwise there is a race condition with other
* processes. */
int fd = open(filename,O_WRONLY|O_CREAT,0644);
if (fd == -1) {
redisLog(REDIS_WARNING,
"Can't open %s in order to acquire a lock: %s",
filename, strerror(errno));
return REDIS_ERR;
}
if (flock(fd,LOCK_EX|LOCK_NB) == -1) {
if (errno == EWOULDBLOCK) {
redisLog(REDIS_WARNING,
"Sorry, the cluster configuration file %s is already used "
"by a different Redis Cluster node. Please make sure that "
"different nodes use different cluster configuration "
"files.", filename);
} else {
redisLog(REDIS_WARNING,
"Impossible to lock %s: %s", filename, strerror(errno));
}
close(fd);
return REDIS_ERR;
}
/* Lock acquired: leak the 'fd' by not closing it, so that we'll retain the
* lock to the file as long as the process exists. */
return REDIS_OK;
}
void clusterInit(void) {
int saveconf = 0;
server.cluster = zmalloc(sizeof(clusterState));
server.cluster->myself = NULL;
server.cluster->currentEpoch = 0;
server.cluster->state = REDIS_CLUSTER_FAIL;
server.cluster->size = 1;
server.cluster->todo_before_sleep = 0;
server.cluster->nodes = dictCreate(&clusterNodesDictType,NULL);
server.cluster->nodes_black_list =
dictCreate(&clusterNodesBlackListDictType,NULL);
server.cluster->failover_auth_time = 0;
server.cluster->failover_auth_count = 0;
server.cluster->failover_auth_rank = 0;
server.cluster->failover_auth_epoch = 0;
server.cluster->lastVoteEpoch = 0;
server.cluster->stats_bus_messages_sent = 0;
server.cluster->stats_bus_messages_received = 0;
memset(server.cluster->slots,0, sizeof(server.cluster->slots));
clusterCloseAllSlots();
/* Lock the cluster config file to make sure every node uses
* its own nodes.conf. */
if (clusterLockConfig(server.cluster_configfile) == REDIS_ERR)
exit(1);
/* Load or create a new nodes configuration. */
if (clusterLoadConfig(server.cluster_configfile) == REDIS_ERR) {
/* No configuration found. We will just use the random name provided
* by the createClusterNode() function. */
myself = server.cluster->myself =
createClusterNode(NULL,REDIS_NODE_MYSELF|REDIS_NODE_MASTER);
redisLog(REDIS_NOTICE,"No cluster configuration found, I'm %.40s",
myself->name);
clusterAddNode(myself);
saveconf = 1;
}
if (saveconf) clusterSaveConfigOrDie(1);
/* We need a listening TCP port for our cluster messaging needs. */
server.cfd_count = 0;
/* Port sanity check II
* The other handshake port check is triggered too late to stop
* us from trying to use a too-high cluster port number. */
if (server.port > (65535-REDIS_CLUSTER_PORT_INCR)) {
redisLog(REDIS_WARNING, "Redis port number too high. "
"Cluster communication port is 10,000 port "
"numbers higher than your Redis port. "
"Your Redis port number must be "
"lower than 55535.");
exit(1);
}
if (listenToPort(server.port+REDIS_CLUSTER_PORT_INCR,
server.cfd,&server.cfd_count) == REDIS_ERR)
{
exit(1);
} else {
int j;
for (j = 0; j < server.cfd_count; j++) {
if (aeCreateFileEvent(server.el, server.cfd[j], AE_READABLE,
clusterAcceptHandler, NULL) == AE_ERR)
redisPanic("Unrecoverable error creating Redis Cluster "
"file event.");
}
}
/* The slots -> keys map is a sorted set. Init it. */
server.cluster->slots_to_keys = zslCreate();
resetManualFailover();
}
/* Reset a node performing a soft or hard reset:
*
* 1) All other nodes are forget.
* 2) All the assigned / open slots are released.
* 3) If the node is a slave, it turns into a master.
* 5) Only for hard reset: a new Node ID is generated.
* 6) Only for hard reset: currentEpoch and configEpoch are set to 0.
* 7) The new configuration is saved and the cluster state updated. */
void clusterReset(int hard) {
dictIterator *di;
dictEntry *de;
int j;
/* Turn into master. */
if (nodeIsSlave(myself)) {
clusterSetNodeAsMaster(myself);
replicationUnsetMaster();
}
/* Close slots, reset manual failover state. */
clusterCloseAllSlots();
resetManualFailover();
/* Unassign all the slots. */
for (j = 0; j < REDIS_CLUSTER_SLOTS; j++) clusterDelSlot(j);
/* Forget all the nodes, but myself. */
di = dictGetSafeIterator(server.cluster->nodes);
while((de = dictNext(di)) != NULL) {
clusterNode *node = dictGetVal(de);
if (node == myself) continue;
clusterDelNode(node);
}
dictReleaseIterator(di);
/* Hard reset only: set epochs to 0, change node ID. */
if (hard) {
sds oldname;
server.cluster->currentEpoch = 0;
server.cluster->lastVoteEpoch = 0;
myself->configEpoch = 0;
redisLog(REDIS_WARNING, "configEpoch set to 0 via CLUSTER RESET HARD");
/* To change the Node ID we need to remove the old name from the
* nodes table, change the ID, and re-add back with new name. */
oldname = sdsnewlen(myself->name, REDIS_CLUSTER_NAMELEN);
dictDelete(server.cluster->nodes,oldname);
sdsfree(oldname);
getRandomHexChars(myself->name, REDIS_CLUSTER_NAMELEN);
clusterAddNode(myself);
}
/* Make sure to persist the new config and update the state. */
clusterDoBeforeSleep(CLUSTER_TODO_SAVE_CONFIG|
CLUSTER_TODO_UPDATE_STATE|
CLUSTER_TODO_FSYNC_CONFIG);
}
/* -----------------------------------------------------------------------------
* CLUSTER communication link
* -------------------------------------------------------------------------- */
clusterLink *createClusterLink(clusterNode *node) {
clusterLink *link = zmalloc(sizeof(*link));
link->ctime = mstime();
link->sndbuf = sdsempty();
link->rcvbuf = sdsempty();
link->node = node;
link->fd = -1;
return link;
}
/* Free a cluster link, but does not free the associated node of course.
* This function will just make sure that the original node associated
* with this link will have the 'link' field set to NULL. */
void freeClusterLink(clusterLink *link) {
if (link->fd != -1) {
aeDeleteFileEvent(server.el, link->fd, AE_WRITABLE);
aeDeleteFileEvent(server.el, link->fd, AE_READABLE);
}
sdsfree(link->sndbuf);
sdsfree(link->rcvbuf);
if (link->node)
link->node->link = NULL;
close(link->fd);
zfree(link);
}
#define MAX_CLUSTER_ACCEPTS_PER_CALL 1000
void clusterAcceptHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
int cport, cfd;
int max = MAX_CLUSTER_ACCEPTS_PER_CALL;
char cip[REDIS_IP_STR_LEN];
clusterLink *link;
REDIS_NOTUSED(el);
REDIS_NOTUSED(mask);
REDIS_NOTUSED(privdata);
/* If the server is starting up, don't accept cluster connections:
* UPDATE messages may interact with the database content. */
if (server.masterhost == NULL && server.loading) return;
while(max--) {
cfd = anetTcpAccept(server.neterr, fd, cip, sizeof(cip), &cport);
if (cfd == ANET_ERR) {
if (errno != EWOULDBLOCK)
redisLog(REDIS_VERBOSE,
"Accepting cluster node: %s", server.neterr);
return;
}
anetNonBlock(NULL,cfd);
anetEnableTcpNoDelay(NULL,cfd);
/* Use non-blocking I/O for cluster messages. */
redisLog(REDIS_VERBOSE,"Accepted cluster node %s:%d", cip, cport);
/* Create a link object we use to handle the connection.
* It gets passed to the readable handler when data is available.
* Initiallly the link->node pointer is set to NULL as we don't know
* which node is, but the right node is references once we know the
* node identity. */
link = createClusterLink(NULL);
link->fd = cfd;
aeCreateFileEvent(server.el,cfd,AE_READABLE,clusterReadHandler,link);
}
}
/* -----------------------------------------------------------------------------
* Key space handling
* -------------------------------------------------------------------------- */
/* We have 16384 hash slots. The hash slot of a given key is obtained
* as the least significant 14 bits of the crc16 of the key.
*
* However if the key contains the {...} pattern, only the part between
* { and } is hashed. This may be useful in the future to force certain
* keys to be in the same node (assuming no resharding is in progress). */
unsigned int keyHashSlot(char *key, int keylen) {
int s, e; /* start-end indexes of { and } */
for (s = 0; s < keylen; s++)
if (key[s] == '{') break;
/* No '{' ? Hash the whole key. This is the base case. */
if (s == keylen) return crc16(key,keylen) & 0x3FFF;
/* '{' found? Check if we have the corresponding '}'. */
for (e = s+1; e < keylen; e++)
if (key[e] == '}') break;
/* No '}' or nothing betweeen {} ? Hash the whole key. */
if (e == keylen || e == s+1) return crc16(key,keylen) & 0x3FFF;
/* If we are here there is both a { and a } on its right. Hash
* what is in the middle between { and }. */
return crc16(key+s+1,e-s-1) & 0x3FFF;
}
/* -----------------------------------------------------------------------------
* CLUSTER node API
* -------------------------------------------------------------------------- */
/* Create a new cluster node, with the specified flags.
* If "nodename" is NULL this is considered a first handshake and a random
* node name is assigned to this node (it will be fixed later when we'll
* receive the first pong).
*
* The node is created and returned to the user, but it is not automatically
* added to the nodes hash table. */
clusterNode *createClusterNode(char *nodename, int flags) {
clusterNode *node = zmalloc(sizeof(*node));
if (nodename)
memcpy(node->name, nodename, REDIS_CLUSTER_NAMELEN);
else
getRandomHexChars(node->name, REDIS_CLUSTER_NAMELEN);
node->ctime = mstime();
node->configEpoch = 0;
node->flags = flags;
memset(node->slots,0,sizeof(node->slots));
node->numslots = 0;
node->numslaves = 0;
node->slaves = NULL;
node->slaveof = NULL;
node->ping_sent = node->pong_received = 0;
node->fail_time = 0;
node->link = NULL;
memset(node->ip,0,sizeof(node->ip));
node->port = 0;
node->fail_reports = listCreate();
node->voted_time = 0;
node->repl_offset_time = 0;
node->repl_offset = 0;
listSetFreeMethod(node->fail_reports,zfree);
return node;
}
/* This function is called every time we get a failure report from a node.
* The side effect is to populate the fail_reports list (or to update
* the timestamp of an existing report).
*
* 'failing' is the node that is in failure state according to the
* 'sender' node.
*
* The function returns 0 if it just updates a timestamp of an existing
* failure report from the same sender. 1 is returned if a new failure
* report is created. */
int clusterNodeAddFailureReport(clusterNode *failing, clusterNode *sender) {
list *l = failing->fail_reports;
listNode *ln;
listIter li;
clusterNodeFailReport *fr;
/* If a failure report from the same sender already exists, just update
* the timestamp. */
listRewind(l,&li);
while ((ln = listNext(&li)) != NULL) {
fr = ln->value;
if (fr->node == sender) {
fr->time = mstime();
return 0;
}
}
/* Otherwise create a new report. */
fr = zmalloc(sizeof(*fr));
fr->node = sender;
fr->time = mstime();
listAddNodeTail(l,fr);
return 1;
}
/* Remove failure reports that are too old, where too old means reasonably
* older than the global node timeout. Note that anyway for a node to be
* flagged as FAIL we need to have a local PFAIL state that is at least
* older than the global node timeout, so we don't just trust the number
* of failure reports from other nodes. */
void clusterNodeCleanupFailureReports(clusterNode *node) {
list *l = node->fail_reports;
listNode *ln;
listIter li;
clusterNodeFailReport *fr;
mstime_t maxtime = server.cluster_node_timeout *
REDIS_CLUSTER_FAIL_REPORT_VALIDITY_MULT;
mstime_t now = mstime();
listRewind(l,&li);
while ((ln = listNext(&li)) != NULL) {
fr = ln->value;
if (now - fr->time > maxtime) listDelNode(l,ln);
}
}
/* Remove the failing report for 'node' if it was previously considered
* failing by 'sender'. This function is called when a node informs us via
* gossip that a node is OK from its point of view (no FAIL or PFAIL flags).
*
* Note that this function is called relatively often as it gets called even
* when there are no nodes failing, and is O(N), however when the cluster is
* fine the failure reports list is empty so the function runs in constant
* time.
*
* The function returns 1 if the failure report was found and removed.
* Otherwise 0 is returned. */
int clusterNodeDelFailureReport(clusterNode *node, clusterNode *sender) {
list *l = node->fail_reports;
listNode *ln;
listIter li;
clusterNodeFailReport *fr;
/* Search for a failure report from this sender. */
listRewind(l,&li);
while ((ln = listNext(&li)) != NULL) {
fr = ln->value;
if (fr->node == sender) break;
}
if (!ln) return 0; /* No failure report from this sender. */
/* Remove the failure report. */
listDelNode(l,ln);
clusterNodeCleanupFailureReports(node);
return 1;
}
/* Return the number of external nodes that believe 'node' is failing,
* not including this node, that may have a PFAIL or FAIL state for this
* node as well. */
int clusterNodeFailureReportsCount(clusterNode *node) {
clusterNodeCleanupFailureReports(node);
return listLength(node->fail_reports);
}
int clusterNodeRemoveSlave(clusterNode *master, clusterNode *slave) {
int j;
for (j = 0; j < master->numslaves; j++) {
if (master->slaves[j] == slave) {
memmove(master->slaves+j,master->slaves+(j+1),
(master->numslaves-1)-j);
master->numslaves--;
return REDIS_OK;
}
}
return REDIS_ERR;
}
int clusterNodeAddSlave(clusterNode *master, clusterNode *slave) {
int j;
/* If it's already a slave, don't add it again. */
for (j = 0; j < master->numslaves; j++)
if (master->slaves[j] == slave) return REDIS_ERR;
master->slaves = zrealloc(master->slaves,
sizeof(clusterNode*)*(master->numslaves+1));
master->slaves[master->numslaves] = slave;
master->numslaves++;
return REDIS_OK;
}
void clusterNodeResetSlaves(clusterNode *n) {
zfree(n->slaves);
n->numslaves = 0;
n->slaves = NULL;
}
int clusterCountNonFailingSlaves(clusterNode *n) {
int j, okslaves = 0;
for (j = 0; j < n->numslaves; j++)
if (!nodeFailed(n->slaves[j])) okslaves++;
return okslaves;
}
void freeClusterNode(clusterNode *n) {
sds nodename;
nodename = sdsnewlen(n->name, REDIS_CLUSTER_NAMELEN);
redisAssert(dictDelete(server.cluster->nodes,nodename) == DICT_OK);
sdsfree(nodename);
if (n->slaveof) clusterNodeRemoveSlave(n->slaveof, n);
if (n->link) freeClusterLink(n->link);
listRelease(n->fail_reports);
zfree(n);
}
/* Add a node to the nodes hash table */
int clusterAddNode(clusterNode *node) {
int retval;
retval = dictAdd(server.cluster->nodes,
sdsnewlen(node->name,REDIS_CLUSTER_NAMELEN), node);
return (retval == DICT_OK) ? REDIS_OK : REDIS_ERR;
}
/* Remove a node from the cluster:
* 1) Mark all the nodes handled by it as unassigned.
* 2) Remove all the failure reports sent by this node.
* 3) Free the node, that will in turn remove it from the hash table
* and from the list of slaves of its master, if it is a slave node.
*/
void clusterDelNode(clusterNode *delnode) {
int j;
dictIterator *di;
dictEntry *de;
/* 1) Mark slots as unassigned. */
for (j = 0; j < REDIS_CLUSTER_SLOTS; j++) {
if (server.cluster->importing_slots_from[j] == delnode)
server.cluster->importing_slots_from[j] = NULL;
if (server.cluster->migrating_slots_to[j] == delnode)
server.cluster->migrating_slots_to[j] = NULL;
if (server.cluster->slots[j] == delnode)
clusterDelSlot(j);
}
/* 2) Remove failure reports. */
di = dictGetSafeIterator(server.cluster->nodes);
while((de = dictNext(di)) != NULL) {
clusterNode *node = dictGetVal(de);
if (node == delnode) continue;
clusterNodeDelFailureReport(node,delnode);
}
dictReleaseIterator(di);
/* 3) Remove this node from its master's slaves if needed. */
if (nodeIsSlave(delnode) && delnode->slaveof)
clusterNodeRemoveSlave(delnode->slaveof,delnode);
/* 4) Free the node, unlinking it from the cluster. */
freeClusterNode(delnode);
}
/* Node lookup by name */
clusterNode *clusterLookupNode(char *name) {
sds s = sdsnewlen(name, REDIS_CLUSTER_NAMELEN);
dictEntry *de;
de = dictFind(server.cluster->nodes,s);
sdsfree(s);
if (de == NULL) return NULL;
return dictGetVal(de);
}
/* This is only used after the handshake. When we connect a given IP/PORT
* as a result of CLUSTER MEET we don't have the node name yet, so we
* pick a random one, and will fix it when we receive the PONG request using
* this function. */
void clusterRenameNode(clusterNode *node, char *newname) {
int retval;
sds s = sdsnewlen(node->name, REDIS_CLUSTER_NAMELEN);
redisLog(REDIS_DEBUG,"Renaming node %.40s into %.40s",
node->name, newname);
retval = dictDelete(server.cluster->nodes, s);
sdsfree(s);
redisAssert(retval == DICT_OK);
memcpy(node->name, newname, REDIS_CLUSTER_NAMELEN);
clusterAddNode(node);
}
/* -----------------------------------------------------------------------------
* CLUSTER nodes blacklist
*
* The nodes blacklist is just a way to ensure that a given node with a given
* Node ID is not readded before some time elapsed (this time is specified
* in seconds in REDIS_CLUSTER_BLACKLIST_TTL).
*
* This is useful when we want to remove a node from the cluster completely:
* when CLUSTER FORGET is called, it also puts the node into the blacklist so
* that even if we receive gossip messages from other nodes that still remember
* about the node we want to remove, we don't re-add it before some time.
*
* Currently the REDIS_CLUSTER_BLACKLIST_TTL is set to 1 minute, this means
* that redis-trib has 60 seconds to send CLUSTER FORGET messages to nodes
* in the cluster without dealing with the problem of other nodes re-adding
* back the node to nodes we already sent the FORGET command to.
*
* The data structure used is a hash table with an sds string representing
* the node ID as key, and the time when it is ok to re-add the node as
* value.
* -------------------------------------------------------------------------- */
#define REDIS_CLUSTER_BLACKLIST_TTL 60 /* 1 minute. */
/* Before of the addNode() or Exists() operations we always remove expired
* entries from the black list. This is an O(N) operation but it is not a
* problem since add / exists operations are called very infrequently and
* the hash table is supposed to contain very little elements at max.
* However without the cleanup during long uptimes and with some automated
* node add/removal procedures, entries could accumulate. */
void clusterBlacklistCleanup(void) {
dictIterator *di;
dictEntry *de;
di = dictGetSafeIterator(server.cluster->nodes_black_list);
while((de = dictNext(di)) != NULL) {
int64_t expire = dictGetUnsignedIntegerVal(de);
if (expire < server.unixtime)
dictDelete(server.cluster->nodes_black_list,dictGetKey(de));
}
dictReleaseIterator(di);
}
/* Cleanup the blacklist and add a new node ID to the black list. */
void clusterBlacklistAddNode(clusterNode *node) {
dictEntry *de;
sds id = sdsnewlen(node->name,REDIS_CLUSTER_NAMELEN);
clusterBlacklistCleanup();
if (dictAdd(server.cluster->nodes_black_list,id,NULL) == DICT_OK) {
/* If the key was added, duplicate the sds string representation of
* the key for the next lookup. We'll free it at the end. */
id = sdsdup(id);
}
de = dictFind(server.cluster->nodes_black_list,id);
dictSetUnsignedIntegerVal(de,time(NULL)+REDIS_CLUSTER_BLACKLIST_TTL);
sdsfree(id);
}
/* Return non-zero if the specified node ID exists in the blacklist.
* You don't need to pass an sds string here, any pointer to 40 bytes
* will work. */
int clusterBlacklistExists(char *nodeid) {
sds id = sdsnewlen(nodeid,REDIS_CLUSTER_NAMELEN);
int retval;
clusterBlacklistCleanup();
retval = dictFind(server.cluster->nodes_black_list,id) != NULL;
sdsfree(id);
return retval;
}
/* -----------------------------------------------------------------------------
* CLUSTER messages exchange - PING/PONG and gossip
* -------------------------------------------------------------------------- */
/* This function checks if a given node should be marked as FAIL.
* It happens if the following conditions are met:
*
* 1) We received enough failure reports from other master nodes via gossip.
* Enough means that the majority of the masters signaled the node is
* down recently.
* 2) We believe this node is in PFAIL state.
*
* If a failure is detected we also inform the whole cluster about this
* event trying to force every other node to set the FAIL flag for the node.
*
* Note that the form of agreement used here is weak, as we collect the majority
* of masters state during some time, and even if we force agreement by
* propagating the FAIL message, because of partitions we may not reach every
* node. However:
*
* 1) Either we reach the majority and eventually the FAIL state will propagate
* to all the cluster.
* 2) Or there is no majority so no slave promotion will be authorized and the
* FAIL flag will be cleared after some time.
*/
void markNodeAsFailingIfNeeded(clusterNode *node) {
int failures;
int needed_quorum = (server.cluster->size / 2) + 1;
if (!nodeTimedOut(node)) return; /* We can reach it. */
if (nodeFailed(node)) return; /* Already FAILing. */
failures = clusterNodeFailureReportsCount(node);