forked from samba-team/samba
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathctdb_call.c
1927 lines (1605 loc) · 53.7 KB
/
ctdb_call.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
/*
ctdb_call protocol code
Copyright (C) Andrew Tridgell 2006
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program; if not, see <http://www.gnu.org/licenses/>.
*/
/*
see http://wiki.samba.org/index.php/Samba_%26_Clustering for
protocol design and packet details
*/
#include "replace.h"
#include "system/network.h"
#include "system/filesys.h"
#include <talloc.h>
#include <tevent.h>
#include "lib/util/dlinklist.h"
#include "lib/util/debug.h"
#include "lib/util/samba_util.h"
#include "lib/util/util_process.h"
#include "ctdb_private.h"
#include "ctdb_client.h"
#include "common/rb_tree.h"
#include "common/reqid.h"
#include "common/system.h"
#include "common/common.h"
#include "common/logging.h"
struct ctdb_sticky_record {
struct ctdb_context *ctdb;
struct ctdb_db_context *ctdb_db;
TDB_CONTEXT *pindown;
};
/*
find the ctdb_db from a db index
*/
struct ctdb_db_context *find_ctdb_db(struct ctdb_context *ctdb, uint32_t id)
{
struct ctdb_db_context *ctdb_db;
for (ctdb_db=ctdb->db_list; ctdb_db; ctdb_db=ctdb_db->next) {
if (ctdb_db->db_id == id) {
break;
}
}
return ctdb_db;
}
/*
a varient of input packet that can be used in lock requeue
*/
static void ctdb_call_input_pkt(void *p, struct ctdb_req_header *hdr)
{
struct ctdb_context *ctdb = talloc_get_type(p, struct ctdb_context);
ctdb_input_pkt(ctdb, hdr);
}
/*
send an error reply
*/
static void ctdb_send_error(struct ctdb_context *ctdb,
struct ctdb_req_header *hdr, uint32_t status,
const char *fmt, ...) PRINTF_ATTRIBUTE(4,5);
static void ctdb_send_error(struct ctdb_context *ctdb,
struct ctdb_req_header *hdr, uint32_t status,
const char *fmt, ...)
{
va_list ap;
struct ctdb_reply_error_old *r;
char *msg;
int msglen, len;
if (ctdb->methods == NULL) {
DEBUG(DEBUG_INFO,(__location__ " Failed to send error. Transport is DOWN\n"));
return;
}
va_start(ap, fmt);
msg = talloc_vasprintf(ctdb, fmt, ap);
if (msg == NULL) {
ctdb_fatal(ctdb, "Unable to allocate error in ctdb_send_error\n");
}
va_end(ap);
msglen = strlen(msg)+1;
len = offsetof(struct ctdb_reply_error_old, msg);
r = ctdb_transport_allocate(ctdb, msg, CTDB_REPLY_ERROR, len + msglen,
struct ctdb_reply_error_old);
CTDB_NO_MEMORY_FATAL(ctdb, r);
r->hdr.destnode = hdr->srcnode;
r->hdr.reqid = hdr->reqid;
r->status = status;
r->msglen = msglen;
memcpy(&r->msg[0], msg, msglen);
ctdb_queue_packet(ctdb, &r->hdr);
talloc_free(msg);
}
/**
* send a redirect reply
*
* The logic behind this function is this:
*
* A client wants to grab a record and sends a CTDB_REQ_CALL packet
* to its local ctdb (ctdb_request_call). If the node is not itself
* the record's DMASTER, it first redirects the packet to the
* record's LMASTER. The LMASTER then redirects the call packet to
* the current DMASTER. Note that this works because of this: When
* a record is migrated off a node, then the new DMASTER is stored
* in the record's copy on the former DMASTER.
*/
static void ctdb_call_send_redirect(struct ctdb_context *ctdb,
struct ctdb_db_context *ctdb_db,
TDB_DATA key,
struct ctdb_req_call_old *c,
struct ctdb_ltdb_header *header)
{
uint32_t lmaster = ctdb_lmaster(ctdb, &key);
c->hdr.destnode = lmaster;
if (ctdb->pnn == lmaster) {
c->hdr.destnode = header->dmaster;
}
c->hopcount++;
if (c->hopcount%100 > 95) {
DEBUG(DEBUG_WARNING,("High hopcount %d dbid:%s "
"key:0x%08x reqid=%08x pnn:%d src:%d lmaster:%d "
"header->dmaster:%d dst:%d\n",
c->hopcount, ctdb_db->db_name, ctdb_hash(&key),
c->hdr.reqid, ctdb->pnn, c->hdr.srcnode, lmaster,
header->dmaster, c->hdr.destnode));
}
ctdb_queue_packet(ctdb, &c->hdr);
}
/*
send a dmaster reply
caller must have the chainlock before calling this routine. Caller must be
the lmaster
*/
static void ctdb_send_dmaster_reply(struct ctdb_db_context *ctdb_db,
struct ctdb_ltdb_header *header,
TDB_DATA key, TDB_DATA data,
uint32_t new_dmaster,
uint32_t reqid)
{
struct ctdb_context *ctdb = ctdb_db->ctdb;
struct ctdb_reply_dmaster_old *r;
int ret, len;
TALLOC_CTX *tmp_ctx;
if (ctdb->pnn != ctdb_lmaster(ctdb, &key)) {
DEBUG(DEBUG_ALERT,(__location__ " Caller is not lmaster!\n"));
return;
}
header->dmaster = new_dmaster;
ret = ctdb_ltdb_store(ctdb_db, key, header, data);
if (ret != 0) {
ctdb_fatal(ctdb, "ctdb_send_dmaster_reply unable to update dmaster");
return;
}
if (ctdb->methods == NULL) {
ctdb_fatal(ctdb, "ctdb_send_dmaster_reply cant update dmaster since transport is down");
return;
}
/* put the packet on a temporary context, allowing us to safely free
it below even if ctdb_reply_dmaster() has freed it already */
tmp_ctx = talloc_new(ctdb);
/* send the CTDB_REPLY_DMASTER */
len = offsetof(struct ctdb_reply_dmaster_old, data) + key.dsize + data.dsize + sizeof(uint32_t);
r = ctdb_transport_allocate(ctdb, tmp_ctx, CTDB_REPLY_DMASTER, len,
struct ctdb_reply_dmaster_old);
CTDB_NO_MEMORY_FATAL(ctdb, r);
r->hdr.destnode = new_dmaster;
r->hdr.reqid = reqid;
r->hdr.generation = ctdb_db->generation;
r->rsn = header->rsn;
r->keylen = key.dsize;
r->datalen = data.dsize;
r->db_id = ctdb_db->db_id;
memcpy(&r->data[0], key.dptr, key.dsize);
memcpy(&r->data[key.dsize], data.dptr, data.dsize);
memcpy(&r->data[key.dsize+data.dsize], &header->flags, sizeof(uint32_t));
ctdb_queue_packet(ctdb, &r->hdr);
talloc_free(tmp_ctx);
}
/*
send a dmaster request (give another node the dmaster for a record)
This is always sent to the lmaster, which ensures that the lmaster
always knows who the dmaster is. The lmaster will then send a
CTDB_REPLY_DMASTER to the new dmaster
*/
static void ctdb_call_send_dmaster(struct ctdb_db_context *ctdb_db,
struct ctdb_req_call_old *c,
struct ctdb_ltdb_header *header,
TDB_DATA *key, TDB_DATA *data)
{
struct ctdb_req_dmaster_old *r;
struct ctdb_context *ctdb = ctdb_db->ctdb;
int len;
uint32_t lmaster = ctdb_lmaster(ctdb, key);
if (ctdb->methods == NULL) {
ctdb_fatal(ctdb, "Failed ctdb_call_send_dmaster since transport is down");
return;
}
if (data->dsize != 0) {
header->flags |= CTDB_REC_FLAG_MIGRATED_WITH_DATA;
}
if (lmaster == ctdb->pnn) {
ctdb_send_dmaster_reply(ctdb_db, header, *key, *data,
c->hdr.srcnode, c->hdr.reqid);
return;
}
len = offsetof(struct ctdb_req_dmaster_old, data) + key->dsize + data->dsize
+ sizeof(uint32_t);
r = ctdb_transport_allocate(ctdb, ctdb, CTDB_REQ_DMASTER, len,
struct ctdb_req_dmaster_old);
CTDB_NO_MEMORY_FATAL(ctdb, r);
r->hdr.destnode = lmaster;
r->hdr.reqid = c->hdr.reqid;
r->hdr.generation = ctdb_db->generation;
r->db_id = c->db_id;
r->rsn = header->rsn;
r->dmaster = c->hdr.srcnode;
r->keylen = key->dsize;
r->datalen = data->dsize;
memcpy(&r->data[0], key->dptr, key->dsize);
memcpy(&r->data[key->dsize], data->dptr, data->dsize);
memcpy(&r->data[key->dsize + data->dsize], &header->flags, sizeof(uint32_t));
header->dmaster = c->hdr.srcnode;
if (ctdb_ltdb_store(ctdb_db, *key, header, *data) != 0) {
ctdb_fatal(ctdb, "Failed to store record in ctdb_call_send_dmaster");
}
ctdb_queue_packet(ctdb, &r->hdr);
talloc_free(r);
}
static void ctdb_sticky_pindown_timeout(struct tevent_context *ev,
struct tevent_timer *te,
struct timeval t, void *private_data)
{
struct ctdb_sticky_record *sr = talloc_get_type(private_data,
struct ctdb_sticky_record);
DEBUG(DEBUG_ERR,("Pindown timeout db:%s unstick record\n", sr->ctdb_db->db_name));
if (sr->pindown != NULL) {
talloc_free(sr->pindown);
sr->pindown = NULL;
}
}
static int
ctdb_set_sticky_pindown(struct ctdb_context *ctdb, struct ctdb_db_context *ctdb_db, TDB_DATA key)
{
TALLOC_CTX *tmp_ctx = talloc_new(NULL);
uint32_t *k;
struct ctdb_sticky_record *sr;
k = ctdb_key_to_idkey(tmp_ctx, key);
if (k == NULL) {
DEBUG(DEBUG_ERR,("Failed to allocate key for sticky record\n"));
talloc_free(tmp_ctx);
return -1;
}
sr = trbt_lookuparray32(ctdb_db->sticky_records, k[0], &k[0]);
if (sr == NULL) {
talloc_free(tmp_ctx);
return 0;
}
talloc_free(tmp_ctx);
if (sr->pindown == NULL) {
DEBUG(DEBUG_ERR,("Pinning down record in %s for %d ms\n", ctdb_db->db_name, ctdb->tunable.sticky_pindown));
sr->pindown = talloc_new(sr);
if (sr->pindown == NULL) {
DEBUG(DEBUG_ERR,("Failed to allocate pindown context for sticky record\n"));
return -1;
}
tevent_add_timer(ctdb->ev, sr->pindown,
timeval_current_ofs(ctdb->tunable.sticky_pindown / 1000,
(ctdb->tunable.sticky_pindown * 1000) % 1000000),
ctdb_sticky_pindown_timeout, sr);
}
return 0;
}
/*
called when a CTDB_REPLY_DMASTER packet comes in, or when the lmaster
gets a CTDB_REQUEST_DMASTER for itself. We become the dmaster.
must be called with the chainlock held. This function releases the chainlock
*/
static void ctdb_become_dmaster(struct ctdb_db_context *ctdb_db,
struct ctdb_req_header *hdr,
TDB_DATA key, TDB_DATA data,
uint64_t rsn, uint32_t record_flags)
{
struct ctdb_call_state *state;
struct ctdb_context *ctdb = ctdb_db->ctdb;
struct ctdb_ltdb_header header;
int ret;
DEBUG(DEBUG_DEBUG,("pnn %u dmaster response %08x\n", ctdb->pnn, ctdb_hash(&key)));
ZERO_STRUCT(header);
header.rsn = rsn;
header.dmaster = ctdb->pnn;
header.flags = record_flags;
state = reqid_find(ctdb->idr, hdr->reqid, struct ctdb_call_state);
if (state) {
if (state->call->flags & CTDB_CALL_FLAG_VACUUM_MIGRATION) {
/*
* We temporarily add the VACUUM_MIGRATED flag to
* the record flags, so that ctdb_ltdb_store can
* decide whether the record should be stored or
* deleted.
*/
header.flags |= CTDB_REC_FLAG_VACUUM_MIGRATED;
}
}
if (ctdb_ltdb_store(ctdb_db, key, &header, data) != 0) {
ctdb_fatal(ctdb, "ctdb_reply_dmaster store failed\n");
ret = ctdb_ltdb_unlock(ctdb_db, key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
return;
}
/* we just became DMASTER and this database is "sticky",
see if the record is flagged as "hot" and set up a pin-down
context to stop migrations for a little while if so
*/
if (ctdb_db->sticky) {
ctdb_set_sticky_pindown(ctdb, ctdb_db, key);
}
if (state == NULL) {
DEBUG(DEBUG_ERR,("pnn %u Invalid reqid %u in ctdb_become_dmaster from node %u\n",
ctdb->pnn, hdr->reqid, hdr->srcnode));
ret = ctdb_ltdb_unlock(ctdb_db, key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
return;
}
if (key.dsize != state->call->key.dsize || memcmp(key.dptr, state->call->key.dptr, key.dsize)) {
DEBUG(DEBUG_ERR, ("Got bogus DMASTER packet reqid:%u from node %u. Key does not match key held in matching idr.\n", hdr->reqid, hdr->srcnode));
ret = ctdb_ltdb_unlock(ctdb_db, key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
return;
}
if (hdr->reqid != state->reqid) {
/* we found a record but it was the wrong one */
DEBUG(DEBUG_ERR, ("Dropped orphan in ctdb_become_dmaster with reqid:%u\n from node %u", hdr->reqid, hdr->srcnode));
ret = ctdb_ltdb_unlock(ctdb_db, key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
return;
}
ctdb_call_local(ctdb_db, state->call, &header, state, &data, true);
ret = ctdb_ltdb_unlock(ctdb_db, state->call->key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
state->state = CTDB_CALL_DONE;
if (state->async.fn) {
state->async.fn(state);
}
}
struct dmaster_defer_call {
struct dmaster_defer_call *next, *prev;
struct ctdb_context *ctdb;
struct ctdb_req_header *hdr;
};
struct dmaster_defer_queue {
struct ctdb_db_context *ctdb_db;
uint32_t generation;
struct dmaster_defer_call *deferred_calls;
};
static void dmaster_defer_reprocess(struct tevent_context *ev,
struct tevent_timer *te,
struct timeval t,
void *private_data)
{
struct dmaster_defer_call *call = talloc_get_type(
private_data, struct dmaster_defer_call);
ctdb_input_pkt(call->ctdb, call->hdr);
talloc_free(call);
}
static int dmaster_defer_queue_destructor(struct dmaster_defer_queue *ddq)
{
/* Ignore requests, if database recovery happens in-between. */
if (ddq->generation != ddq->ctdb_db->generation) {
return 0;
}
while (ddq->deferred_calls != NULL) {
struct dmaster_defer_call *call = ddq->deferred_calls;
DLIST_REMOVE(ddq->deferred_calls, call);
talloc_steal(call->ctdb, call);
tevent_add_timer(call->ctdb->ev, call, timeval_zero(),
dmaster_defer_reprocess, call);
}
return 0;
}
static void *insert_ddq_callback(void *parm, void *data)
{
if (data) {
talloc_free(data);
}
return parm;
}
/**
* This function is used to reigster a key in database that needs to be updated.
* Any requests for that key should get deferred till this is completed.
*/
static int dmaster_defer_setup(struct ctdb_db_context *ctdb_db,
struct ctdb_req_header *hdr,
TDB_DATA key)
{
uint32_t *k;
struct dmaster_defer_queue *ddq;
k = ctdb_key_to_idkey(hdr, key);
if (k == NULL) {
DEBUG(DEBUG_ERR, ("Failed to allocate key for dmaster defer setup\n"));
return -1;
}
/* Already exists */
ddq = trbt_lookuparray32(ctdb_db->defer_dmaster, k[0], k);
if (ddq != NULL) {
if (ddq->generation == ctdb_db->generation) {
talloc_free(k);
return 0;
}
/* Recovery ocurred - get rid of old queue. All the deferred
* requests will be resent anyway from ctdb_call_resend_db.
*/
talloc_free(ddq);
}
ddq = talloc(hdr, struct dmaster_defer_queue);
if (ddq == NULL) {
DEBUG(DEBUG_ERR, ("Failed to allocate dmaster defer queue\n"));
talloc_free(k);
return -1;
}
ddq->ctdb_db = ctdb_db;
ddq->generation = hdr->generation;
ddq->deferred_calls = NULL;
trbt_insertarray32_callback(ctdb_db->defer_dmaster, k[0], k,
insert_ddq_callback, ddq);
talloc_set_destructor(ddq, dmaster_defer_queue_destructor);
talloc_free(k);
return 0;
}
static int dmaster_defer_add(struct ctdb_db_context *ctdb_db,
struct ctdb_req_header *hdr,
TDB_DATA key)
{
struct dmaster_defer_queue *ddq;
struct dmaster_defer_call *call;
uint32_t *k;
k = ctdb_key_to_idkey(hdr, key);
if (k == NULL) {
DEBUG(DEBUG_ERR, ("Failed to allocate key for dmaster defer add\n"));
return -1;
}
ddq = trbt_lookuparray32(ctdb_db->defer_dmaster, k[0], k);
if (ddq == NULL) {
talloc_free(k);
return -1;
}
talloc_free(k);
if (ddq->generation != hdr->generation) {
talloc_set_destructor(ddq, NULL);
talloc_free(ddq);
return -1;
}
call = talloc(ddq, struct dmaster_defer_call);
if (call == NULL) {
DEBUG(DEBUG_ERR, ("Failed to allocate dmaster defer call\n"));
return -1;
}
call->ctdb = ctdb_db->ctdb;
call->hdr = talloc_steal(call, hdr);
DLIST_ADD_END(ddq->deferred_calls, call);
return 0;
}
/*
called when a CTDB_REQ_DMASTER packet comes in
this comes into the lmaster for a record when the current dmaster
wants to give up the dmaster role and give it to someone else
*/
void ctdb_request_dmaster(struct ctdb_context *ctdb, struct ctdb_req_header *hdr)
{
struct ctdb_req_dmaster_old *c = (struct ctdb_req_dmaster_old *)hdr;
TDB_DATA key, data, data2;
struct ctdb_ltdb_header header;
struct ctdb_db_context *ctdb_db;
uint32_t record_flags = 0;
size_t len;
int ret;
key.dptr = c->data;
key.dsize = c->keylen;
data.dptr = c->data + c->keylen;
data.dsize = c->datalen;
len = offsetof(struct ctdb_req_dmaster_old, data) + key.dsize + data.dsize
+ sizeof(uint32_t);
if (len <= c->hdr.length) {
memcpy(&record_flags, &c->data[c->keylen + c->datalen],
sizeof(record_flags));
}
ctdb_db = find_ctdb_db(ctdb, c->db_id);
if (!ctdb_db) {
ctdb_send_error(ctdb, hdr, -1,
"Unknown database in request. db_id==0x%08x",
c->db_id);
return;
}
dmaster_defer_setup(ctdb_db, hdr, key);
/* fetch the current record */
ret = ctdb_ltdb_lock_fetch_requeue(ctdb_db, key, &header, hdr, &data2,
ctdb_call_input_pkt, ctdb, false);
if (ret == -1) {
ctdb_fatal(ctdb, "ctdb_req_dmaster failed to fetch record");
return;
}
if (ret == -2) {
DEBUG(DEBUG_INFO,(__location__ " deferring ctdb_request_dmaster\n"));
return;
}
if (ctdb_lmaster(ctdb, &key) != ctdb->pnn) {
DEBUG(DEBUG_ERR, ("dmaster request to non-lmaster "
"db=%s lmaster=%u gen=%u curgen=%u\n",
ctdb_db->db_name, ctdb_lmaster(ctdb, &key),
hdr->generation, ctdb_db->generation));
ctdb_fatal(ctdb, "ctdb_req_dmaster to non-lmaster");
}
DEBUG(DEBUG_DEBUG,("pnn %u dmaster request on %08x for %u from %u\n",
ctdb->pnn, ctdb_hash(&key), c->dmaster, c->hdr.srcnode));
/* its a protocol error if the sending node is not the current dmaster */
if (header.dmaster != hdr->srcnode) {
DEBUG(DEBUG_ALERT,("pnn %u dmaster request for new-dmaster %u from non-master %u real-dmaster=%u key %08x dbid 0x%08x gen=%u curgen=%u c->rsn=%llu header.rsn=%llu reqid=%u keyval=0x%08x\n",
ctdb->pnn, c->dmaster, hdr->srcnode, header.dmaster, ctdb_hash(&key),
ctdb_db->db_id, hdr->generation, ctdb->vnn_map->generation,
(unsigned long long)c->rsn, (unsigned long long)header.rsn, c->hdr.reqid,
(key.dsize >= 4)?(*(uint32_t *)key.dptr):0));
if (header.rsn != 0 || header.dmaster != ctdb->pnn) {
DEBUG(DEBUG_ERR,("ctdb_req_dmaster from non-master. Force a recovery.\n"));
ctdb->recovery_mode = CTDB_RECOVERY_ACTIVE;
ctdb_ltdb_unlock(ctdb_db, key);
return;
}
}
if (header.rsn > c->rsn) {
DEBUG(DEBUG_ALERT,("pnn %u dmaster request with older RSN new-dmaster %u from %u real-dmaster=%u key %08x dbid 0x%08x gen=%u curgen=%u c->rsn=%llu header.rsn=%llu reqid=%u\n",
ctdb->pnn, c->dmaster, hdr->srcnode, header.dmaster, ctdb_hash(&key),
ctdb_db->db_id, hdr->generation, ctdb->vnn_map->generation,
(unsigned long long)c->rsn, (unsigned long long)header.rsn, c->hdr.reqid));
}
/* use the rsn from the sending node */
header.rsn = c->rsn;
/* store the record flags from the sending node */
header.flags = record_flags;
/* check if the new dmaster is the lmaster, in which case we
skip the dmaster reply */
if (c->dmaster == ctdb->pnn) {
ctdb_become_dmaster(ctdb_db, hdr, key, data, c->rsn, record_flags);
} else {
ctdb_send_dmaster_reply(ctdb_db, &header, key, data, c->dmaster, hdr->reqid);
ret = ctdb_ltdb_unlock(ctdb_db, key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
}
}
static void ctdb_sticky_record_timeout(struct tevent_context *ev,
struct tevent_timer *te,
struct timeval t, void *private_data)
{
struct ctdb_sticky_record *sr = talloc_get_type(private_data,
struct ctdb_sticky_record);
talloc_free(sr);
}
static void *ctdb_make_sticky_record_callback(void *parm, void *data)
{
if (data) {
DEBUG(DEBUG_ERR,("Already have sticky record registered. Free old %p and create new %p\n", data, parm));
talloc_free(data);
}
return parm;
}
static int
ctdb_make_record_sticky(struct ctdb_context *ctdb, struct ctdb_db_context *ctdb_db, TDB_DATA key)
{
TALLOC_CTX *tmp_ctx = talloc_new(NULL);
uint32_t *k;
struct ctdb_sticky_record *sr;
k = ctdb_key_to_idkey(tmp_ctx, key);
if (k == NULL) {
DEBUG(DEBUG_ERR,("Failed to allocate key for sticky record\n"));
talloc_free(tmp_ctx);
return -1;
}
sr = trbt_lookuparray32(ctdb_db->sticky_records, k[0], &k[0]);
if (sr != NULL) {
talloc_free(tmp_ctx);
return 0;
}
sr = talloc(ctdb_db->sticky_records, struct ctdb_sticky_record);
if (sr == NULL) {
talloc_free(tmp_ctx);
DEBUG(DEBUG_ERR,("Failed to allocate sticky record structure\n"));
return -1;
}
sr->ctdb = ctdb;
sr->ctdb_db = ctdb_db;
sr->pindown = NULL;
DEBUG(DEBUG_ERR,("Make record sticky for %d seconds in db %s key:0x%08x.\n",
ctdb->tunable.sticky_duration,
ctdb_db->db_name, ctdb_hash(&key)));
trbt_insertarray32_callback(ctdb_db->sticky_records, k[0], &k[0], ctdb_make_sticky_record_callback, sr);
tevent_add_timer(ctdb->ev, sr,
timeval_current_ofs(ctdb->tunable.sticky_duration, 0),
ctdb_sticky_record_timeout, sr);
talloc_free(tmp_ctx);
return 0;
}
struct pinned_down_requeue_handle {
struct ctdb_context *ctdb;
struct ctdb_req_header *hdr;
};
struct pinned_down_deferred_call {
struct ctdb_context *ctdb;
struct ctdb_req_header *hdr;
};
static void pinned_down_requeue(struct tevent_context *ev,
struct tevent_timer *te,
struct timeval t, void *private_data)
{
struct pinned_down_requeue_handle *handle = talloc_get_type(private_data, struct pinned_down_requeue_handle);
struct ctdb_context *ctdb = handle->ctdb;
talloc_steal(ctdb, handle->hdr);
ctdb_call_input_pkt(ctdb, handle->hdr);
talloc_free(handle);
}
static int pinned_down_destructor(struct pinned_down_deferred_call *pinned_down)
{
struct ctdb_context *ctdb = pinned_down->ctdb;
struct pinned_down_requeue_handle *handle = talloc(ctdb, struct pinned_down_requeue_handle);
handle->ctdb = pinned_down->ctdb;
handle->hdr = pinned_down->hdr;
talloc_steal(handle, handle->hdr);
tevent_add_timer(ctdb->ev, handle, timeval_zero(),
pinned_down_requeue, handle);
return 0;
}
static int
ctdb_defer_pinned_down_request(struct ctdb_context *ctdb, struct ctdb_db_context *ctdb_db, TDB_DATA key, struct ctdb_req_header *hdr)
{
TALLOC_CTX *tmp_ctx = talloc_new(NULL);
uint32_t *k;
struct ctdb_sticky_record *sr;
struct pinned_down_deferred_call *pinned_down;
k = ctdb_key_to_idkey(tmp_ctx, key);
if (k == NULL) {
DEBUG(DEBUG_ERR,("Failed to allocate key for sticky record\n"));
talloc_free(tmp_ctx);
return -1;
}
sr = trbt_lookuparray32(ctdb_db->sticky_records, k[0], &k[0]);
if (sr == NULL) {
talloc_free(tmp_ctx);
return -1;
}
talloc_free(tmp_ctx);
if (sr->pindown == NULL) {
return -1;
}
pinned_down = talloc(sr->pindown, struct pinned_down_deferred_call);
if (pinned_down == NULL) {
DEBUG(DEBUG_ERR,("Failed to allocate structure for deferred pinned down request\n"));
return -1;
}
pinned_down->ctdb = ctdb;
pinned_down->hdr = hdr;
talloc_set_destructor(pinned_down, pinned_down_destructor);
talloc_steal(pinned_down, hdr);
return 0;
}
static void
ctdb_update_db_stat_hot_keys(struct ctdb_db_context *ctdb_db, TDB_DATA key, int hopcount)
{
int i, id;
/* smallest value is always at index 0 */
if (hopcount <= ctdb_db->statistics.hot_keys[0].count) {
return;
}
/* see if we already know this key */
for (i = 0; i < MAX_HOT_KEYS; i++) {
if (key.dsize != ctdb_db->statistics.hot_keys[i].key.dsize) {
continue;
}
if (memcmp(key.dptr, ctdb_db->statistics.hot_keys[i].key.dptr, key.dsize)) {
continue;
}
/* found an entry for this key */
if (hopcount <= ctdb_db->statistics.hot_keys[i].count) {
return;
}
ctdb_db->statistics.hot_keys[i].count = hopcount;
goto sort_keys;
}
if (ctdb_db->statistics.num_hot_keys < MAX_HOT_KEYS) {
id = ctdb_db->statistics.num_hot_keys;
ctdb_db->statistics.num_hot_keys++;
} else {
id = 0;
}
if (ctdb_db->statistics.hot_keys[id].key.dptr != NULL) {
talloc_free(ctdb_db->statistics.hot_keys[id].key.dptr);
}
ctdb_db->statistics.hot_keys[id].key.dsize = key.dsize;
ctdb_db->statistics.hot_keys[id].key.dptr = talloc_memdup(ctdb_db, key.dptr, key.dsize);
ctdb_db->statistics.hot_keys[id].count = hopcount;
DEBUG(DEBUG_NOTICE,("Updated hot key database=%s key=0x%08x id=%d hop_count=%d\n",
ctdb_db->db_name, ctdb_hash(&key), id, hopcount));
sort_keys:
for (i = 1; i < MAX_HOT_KEYS; i++) {
if (ctdb_db->statistics.hot_keys[i].count == 0) {
continue;
}
if (ctdb_db->statistics.hot_keys[i].count < ctdb_db->statistics.hot_keys[0].count) {
hopcount = ctdb_db->statistics.hot_keys[i].count;
ctdb_db->statistics.hot_keys[i].count = ctdb_db->statistics.hot_keys[0].count;
ctdb_db->statistics.hot_keys[0].count = hopcount;
key = ctdb_db->statistics.hot_keys[i].key;
ctdb_db->statistics.hot_keys[i].key = ctdb_db->statistics.hot_keys[0].key;
ctdb_db->statistics.hot_keys[0].key = key;
}
}
}
/*
called when a CTDB_REQ_CALL packet comes in
*/
void ctdb_request_call(struct ctdb_context *ctdb, struct ctdb_req_header *hdr)
{
struct ctdb_req_call_old *c = (struct ctdb_req_call_old *)hdr;
TDB_DATA data;
struct ctdb_reply_call_old *r;
int ret, len;
struct ctdb_ltdb_header header;
struct ctdb_call *call;
struct ctdb_db_context *ctdb_db;
int tmp_count, bucket;
if (ctdb->methods == NULL) {
DEBUG(DEBUG_INFO,(__location__ " Failed ctdb_request_call. Transport is DOWN\n"));
return;
}
ctdb_db = find_ctdb_db(ctdb, c->db_id);
if (!ctdb_db) {
ctdb_send_error(ctdb, hdr, -1,
"Unknown database in request. db_id==0x%08x",
c->db_id);
return;
}
call = talloc(hdr, struct ctdb_call);
CTDB_NO_MEMORY_FATAL(ctdb, call);
call->call_id = c->callid;
call->key.dptr = c->data;
call->key.dsize = c->keylen;
call->call_data.dptr = c->data + c->keylen;
call->call_data.dsize = c->calldatalen;
call->reply_data.dptr = NULL;
call->reply_data.dsize = 0;
/* If this record is pinned down we should defer the
request until the pindown times out
*/
if (ctdb_db->sticky) {
if (ctdb_defer_pinned_down_request(ctdb, ctdb_db, call->key, hdr) == 0) {
DEBUG(DEBUG_WARNING,
("Defer request for pinned down record in %s\n", ctdb_db->db_name));
talloc_free(call);
return;
}
}
if (dmaster_defer_add(ctdb_db, hdr, call->key) == 0) {
talloc_free(call);
return;
}
/* determine if we are the dmaster for this key. This also
fetches the record data (if any), thus avoiding a 2nd fetch of the data
if the call will be answered locally */
ret = ctdb_ltdb_lock_fetch_requeue(ctdb_db, call->key, &header, hdr, &data,
ctdb_call_input_pkt, ctdb, false);
if (ret == -1) {
ctdb_send_error(ctdb, hdr, ret, "ltdb fetch failed in ctdb_request_call");
talloc_free(call);
return;
}
if (ret == -2) {
DEBUG(DEBUG_INFO,(__location__ " deferred ctdb_request_call\n"));
talloc_free(call);
return;
}
/* Dont do READONLY if we don't have a tracking database */
if ((c->flags & CTDB_WANT_READONLY) && !ctdb_db->readonly) {
c->flags &= ~CTDB_WANT_READONLY;
}
if (header.flags & CTDB_REC_RO_REVOKE_COMPLETE) {
header.flags &= ~CTDB_REC_RO_FLAGS;
CTDB_INCREMENT_STAT(ctdb, total_ro_revokes);
CTDB_INCREMENT_DB_STAT(ctdb_db, db_ro_revokes);
if (ctdb_ltdb_store(ctdb_db, call->key, &header, data) != 0) {
ctdb_fatal(ctdb, "Failed to write header with cleared REVOKE flag");
}
/* and clear out the tracking data */
if (tdb_delete(ctdb_db->rottdb, call->key) != 0) {
DEBUG(DEBUG_ERR,(__location__ " Failed to clear out trackingdb record\n"));
}
}
/* if we are revoking, we must defer all other calls until the revoke
* had completed.
*/
if (header.flags & CTDB_REC_RO_REVOKING_READONLY) {
talloc_free(data.dptr);
ret = ctdb_ltdb_unlock(ctdb_db, call->key);
if (ctdb_add_revoke_deferred_call(ctdb, ctdb_db, call->key, hdr, ctdb_call_input_pkt, ctdb) != 0) {
ctdb_fatal(ctdb, "Failed to add deferred call for revoke child");
}
talloc_free(call);
return;
}
/*
* If we are not the dmaster and are not hosting any delegations,
* then we redirect the request to the node than can answer it
* (the lmaster or the dmaster).
*/
if ((header.dmaster != ctdb->pnn)
&& (!(header.flags & CTDB_REC_RO_HAVE_DELEGATIONS)) ) {
talloc_free(data.dptr);
ctdb_call_send_redirect(ctdb, ctdb_db, call->key, c, &header);
ret = ctdb_ltdb_unlock(ctdb_db, call->key);
if (ret != 0) {
DEBUG(DEBUG_ERR,(__location__ " ctdb_ltdb_unlock() failed with error %d\n", ret));
}
talloc_free(call);
return;
}