-
Notifications
You must be signed in to change notification settings - Fork 390
/
flow_stats_collector.cc
1256 lines (1123 loc) · 47.2 KB
/
flow_stats_collector.cc
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
/*
* Copyright (c) 2013 Juniper Networks, Inc. All rights reserved.
*/
#include <bitset>
#include <boost/uuid/uuid.hpp>
#include <boost/uuid/uuid_io.hpp>
#include <db/db.h>
#include <base/util.h>
#include <base/string_util.h>
#include <cmn/agent_cmn.h>
#include <init/agent_param.h>
#include <boost/functional/factory.hpp>
#include <cmn/agent_factory.h>
#include <oper/interface_common.h>
#include <oper/mirror_table.h>
#include <oper/global_vrouter.h>
#include <ksync/ksync_index.h>
#include <ksync/ksync_entry.h>
#include <ksync/ksync_object.h>
#include <ksync/ksync_netlink.h>
#include <ksync/ksync_sock.h>
#include <uve/agent_uve.h>
#include <vrouter/flow_stats/flow_stats_collector.h>
#include <uve/vn_uve_table.h>
#include <uve/vm_uve_table.h>
#include <uve/interface_uve_stats_table.h>
#include <uve/vrouter_uve_entry.h>
#include <algorithm>
#include <pkt/flow_proto.h>
#include <pkt/flow_mgmt.h>
#include <vrouter/ksync/ksync_init.h>
#include <vrouter/flow_stats/flow_stats_types.h>
FlowStatsCollector::FlowStatsCollector(boost::asio::io_service &io, int intvl,
uint32_t flow_cache_timeout,
AgentUveBase *uve,
uint32_t instance_id,
FlowAgingTableKey *key,
FlowStatsManager *aging_module) :
StatsCollector(TaskScheduler::GetInstance()->GetTaskId
(kTaskFlowStatsCollector), instance_id,
io, kFlowStatsTimerInterval, "Flow stats collector"),
agent_uve_(uve),
task_id_(uve->agent()->task_scheduler()->GetTaskId
(kTaskFlowStatsCollector)),
rand_gen_(boost::uuids::random_generator()),
flow_iteration_key_(NULL),
entries_to_visit_(0),
flow_tcp_syn_age_time_(FlowTcpSynAgeTime),
request_queue_(agent_uve_->agent()->task_scheduler()->
GetTaskId(kTaskFlowStatsCollector),
instance_id,
boost::bind(&FlowStatsCollector::RequestHandler,
this, _1)),
msg_list_(kMaxFlowMsgsPerSend, FlowLogData()), msg_index_(0),
flow_aging_key_(*key), instance_id_(instance_id),
flow_stats_manager_(aging_module), ageing_task_(NULL) {
if (flow_cache_timeout) {
// Convert to usec
flow_age_time_intvl_ = 1000000L * (uint64_t)flow_cache_timeout;
} else {
flow_age_time_intvl_ = FlowAgeTime;
}
deleted_ = false;
request_queue_.set_name("Flow stats collector");
request_queue_.set_measure_busy_time
(agent_uve_->agent()->MeasureQueueDelay());
// Aging timer fires every kFlowStatsTimerInterval msec. Compute
// number of timer fires needed to scan complete table
timers_per_scan_ = TimersPerScan();
}
FlowStatsCollector::~FlowStatsCollector() {
flow_stats_manager_->FreeIndex(instance_id_);
}
boost::uuids::uuid FlowStatsCollector::rand_gen() {
return rand_gen_();
}
uint64_t FlowStatsCollector::GetCurrentTime() {
return UTCTimestampUsec();
}
void FlowStatsCollector::Shutdown() {
assert(ageing_task_ == NULL);
StatsCollector::Shutdown();
request_queue_.Shutdown();
}
// We want to scan the flow table every 25% of configured ageing time.
// Compute number of timer fires needed to scan the flow-table once.
uint32_t FlowStatsCollector::TimersPerScan() {
uint64_t scan_time_millisec;
/* Use Age Time itself as scan-time for flows */
// Convert aging-time configured in micro-sec to millisecond
scan_time_millisec = flow_age_time_intvl_ / 1000;
// Compute time in which we must scan the complete table to honor the
// kFlowScanTime
scan_time_millisec = (scan_time_millisec * kFlowScanTime) / 100;
// Enforce min value on scan-time
if (scan_time_millisec < kFlowStatsTimerInterval) {
scan_time_millisec = kFlowStatsTimerInterval;
}
// Number of timer fires needed to scan table once
return scan_time_millisec / kFlowStatsTimerInterval;
}
// Update entries_to_visit_ based on total flows
// Timer fires every kFlowScanTime. Its possible that we may not have visited
// all entries by the time next timer fires. So, keep accumulating the number
// of entries to visit into entries_to_visit_
//
// A lower-bound and an upper-bound are enforced on entries_to_visit_
void FlowStatsCollector::UpdateEntriesToVisit() {
// Compute number of flows to visit per scan-time
uint32_t entries = flow_tree_.size() / timers_per_scan_;
// Update number of entries to visit in flow.
// The scan for previous timer may still be in progress. So, accmulate
// number of entries to visit
entries_to_visit_ += entries;
// Cap number of entries to visit to 25% of table
if (entries_to_visit_ > ((flow_tree_.size() * kFlowScanTime)/100))
entries_to_visit_ = (flow_tree_.size() * kFlowScanTime)/100;
// Apply lower-limit
if (entries_to_visit_ < kMinFlowsPerTimer)
entries_to_visit_ = kMinFlowsPerTimer;
return;
}
bool FlowStatsCollector::ShouldBeAged(FlowExportInfo *info,
const vr_flow_entry *k_flow,
uint64_t curr_time) {
FlowEntry *flow = info->flow();
//If both forward and reverse flow are marked
//as TCP closed then immediately remote the flow
if (k_flow != NULL) {
uint64_t k_flow_bytes, bytes;
k_flow_bytes = GetFlowStats(k_flow->fe_stats.flow_bytes_oflow,
k_flow->fe_stats.flow_bytes);
bytes = 0x0000ffffffffffffULL & info->bytes();
/* Don't account for agent overflow bits while comparing change in
* stats */
if (bytes < k_flow_bytes) {
return false;
}
}
uint64_t diff_time = curr_time - info->last_modified_time();
if (diff_time < flow_age_time_intvl()) {
return false;
}
if (flow->is_flags_set(FlowEntry::BgpRouterService)) {
return false;
}
return true;
}
uint64_t FlowStatsCollector::GetFlowStats(const uint16_t &oflow_data,
const uint32_t &data) {
uint64_t flow_stats = (uint64_t) oflow_data << (sizeof(uint32_t) * 8);
flow_stats |= data;
return flow_stats;
}
uint64_t FlowStatsCollector::GetUpdatedFlowBytes(const FlowExportInfo *stats,
uint64_t k_flow_bytes) {
uint64_t oflow_bytes = 0xffff000000000000ULL & stats->bytes();
uint64_t old_bytes = 0x0000ffffffffffffULL & stats->bytes();
if (old_bytes > k_flow_bytes) {
oflow_bytes += 0x0001000000000000ULL;
}
return (oflow_bytes |= k_flow_bytes);
}
uint64_t FlowStatsCollector::GetUpdatedFlowPackets(const FlowExportInfo *stats,
uint64_t k_flow_pkts) {
uint64_t oflow_pkts = 0xffffff0000000000ULL & stats->packets();
uint64_t old_pkts = 0x000000ffffffffffULL & stats->packets();
if (old_pkts > k_flow_pkts) {
oflow_pkts += 0x0000010000000000ULL;
}
return (oflow_pkts |= k_flow_pkts);
}
void FlowStatsCollector::UpdateFloatingIpStats(const FlowExportInfo *flow,
uint64_t bytes, uint64_t pkts) {
InterfaceUveTable::FipInfo fip_info;
FlowEntry *fe = flow->flow();
if (!fe) {
return;
}
/* Ignore Non-Floating-IP flow */
if (!fe->fip() || fe->fip_vmi().uuid_ == nil_uuid()) {
return;
}
InterfaceUveStatsTable *table = static_cast<InterfaceUveStatsTable *>
(agent_uve_->interface_uve_table());
fip_info.bytes_ = bytes;
fip_info.packets_ = pkts;
fip_info.fip_ = fe->fip();
fip_info.fip_vmi_ = fe->fip_vmi();
fip_info.is_local_flow_ = fe->is_flags_set(FlowEntry::LocalFlow);
fip_info.is_ingress_flow_ = fe->is_flags_set(FlowEntry::IngressDir);
fip_info.is_reverse_flow_ = fe->is_flags_set(FlowEntry::ReverseFlow);
fip_info.vn_ = fe->data().source_vn_match;
fip_info.rev_fip_ = NULL;
if (fe->fip() != ReverseFlowFip(flow)) {
/* This is the case where Source and Destination VMs (part of
* same compute node) ping to each other to their respective
* Floating IPs. In this case for each flow we need to increment
* stats for both the VMs */
fip_info.rev_fip_ = ReverseFlowFipEntry(flow);
}
table->UpdateFloatingIpStats(fip_info);
}
InterfaceUveTable::FloatingIp *FlowStatsCollector::ReverseFlowFipEntry
(const FlowExportInfo *flow) {
uint32_t fip = ReverseFlowFip(flow);
VmInterfaceKey vmi = ReverseFlowFipVmi(flow);
Interface *intf = dynamic_cast<Interface *>
(agent_uve_->agent()->interface_table()->FindActiveEntry(&vmi));
if (intf) {
InterfaceUveStatsTable *table = static_cast<InterfaceUveStatsTable *>
(agent_uve_->interface_uve_table());
const string &vn = flow->flow()->data().source_vn_match;
return table->FipEntry(fip, vn, intf);
}
return NULL;
}
uint32_t FlowStatsCollector::ReverseFlowFip(const FlowExportInfo *info) {
FlowEntry *rflow = info->reverse_flow();
if (rflow) {
return rflow->fip();
}
return 0;
}
VmInterfaceKey FlowStatsCollector::ReverseFlowFipVmi
(const FlowExportInfo *info)
{
FlowEntry *rflow = info->reverse_flow();
if (rflow) {
return rflow->fip_vmi();
}
return VmInterfaceKey(AgentKey::ADD_DEL_CHANGE, nil_uuid(), "");
}
void FlowStatsCollector::UpdateInterVnStats(FlowExportInfo *info,
uint64_t bytes, uint64_t pkts) {
FlowEntry *flow = info->flow();
string src_vn = flow->data().source_vn_match;
string dst_vn = flow->data().dest_vn_match;
VnUveTable *vn_table = static_cast<VnUveTable *>
(agent_uve_->vn_uve_table());
if (!src_vn.length())
src_vn = FlowHandler::UnknownVn();
if (!dst_vn.length())
dst_vn = FlowHandler::UnknownVn();
/* When packet is going from src_vn to dst_vn it should be interpreted
* as ingress to vrouter and hence in-stats for src_vn w.r.t. dst_vn
* should be incremented. Similarly when the packet is egressing vrouter
* it should be considered as out-stats for dst_vn w.r.t. src_vn.
* Here the direction "in" and "out" should be interpreted w.r.t vrouter
*/
if (flow->is_flags_set(FlowEntry::LocalFlow)) {
vn_table->UpdateInterVnStats(src_vn, dst_vn, bytes, pkts, false);
vn_table->UpdateInterVnStats(dst_vn, src_vn, bytes, pkts, true);
} else {
if (flow->is_flags_set(FlowEntry::IngressDir)) {
vn_table->UpdateInterVnStats(src_vn, dst_vn, bytes, pkts, false);
} else {
vn_table->UpdateInterVnStats(dst_vn, src_vn, bytes, pkts, true);
}
}
}
void FlowStatsCollector::UpdateStatsAndExportFlow(FlowExportInfo *info,
uint64_t teardown_time,
const RevFlowDepParams *p) {
KSyncFlowMemory *ksync_obj = agent_uve_->agent()->ksync()->
ksync_flow_memory();
if (!info) {
return;
}
FlowEntry *fe = info->flow();
const vr_flow_entry *k_flow = ksync_obj->GetValidKFlowEntry(fe->key(),
fe->flow_handle(),
fe->gen_id());
if (k_flow) {
UpdateAndExportInternal(info, k_flow->fe_stats.flow_bytes,
k_flow->fe_stats.flow_bytes_oflow,
k_flow->fe_stats.flow_packets,
k_flow->fe_stats.flow_packets_oflow,
teardown_time, true, p);
return;
}
/* If reading of stats fails, send a message with just teardown time */
info->set_teardown_time(teardown_time);
ExportFlow(info, 0, 0, p);
}
void FlowStatsCollector::FlowDeleteEnqueue(FlowExportInfo *info, uint64_t t) {
FlowEntry *fe = info->flow();
agent_uve_->agent()->pkt()->get_flow_proto()->DeleteFlowRequest(fe);
info->set_delete_enqueue_time(t);
FlowEntry *rflow = info->reverse_flow();
if (rflow) {
FlowExportInfo *rev_info = FindFlowExportInfo(rflow);
if (rev_info) {
rev_info->set_delete_enqueue_time(t);
}
}
}
void FlowStatsCollector::FlowEvictEnqueue(FlowExportInfo *info, uint64_t t,
uint32_t flow_handle,
uint16_t gen_id) {
FlowEntry *fe = info->flow();
agent_uve_->agent()->pkt()->get_flow_proto()->EvictFlowRequest
(fe, flow_handle, gen_id, (gen_id + 1));
info->set_evict_enqueue_time(t);
}
void FlowStatsCollector::UpdateFlowStatsInternal(FlowExportInfo *info,
uint32_t bytes,
uint16_t oflow_bytes,
uint32_t pkts,
uint16_t oflow_pkts,
uint64_t time,
bool teardown_time,
uint64_t *diff_bytes,
uint64_t *diff_pkts) {
uint64_t k_bytes, k_packets, total_bytes, total_packets;
k_bytes = GetFlowStats(oflow_bytes, bytes);
k_packets = GetFlowStats(oflow_pkts, pkts);
total_bytes = GetUpdatedFlowBytes(info, k_bytes);
total_packets = GetUpdatedFlowPackets(info, k_packets);
*diff_bytes = total_bytes - info->bytes();
*diff_pkts = total_packets - info->packets();
info->set_bytes(total_bytes);
info->set_packets(total_packets);
//Update Inter-VN stats
UpdateInterVnStats(info, *diff_bytes, *diff_pkts);
//Update Floating-IP stats
UpdateFloatingIpStats(info, *diff_bytes, *diff_pkts);
if (teardown_time) {
info->set_teardown_time(time);
} else {
info->set_last_modified_time(time);
}
}
void FlowStatsCollector::UpdateAndExportInternalLocked(FlowExportInfo *info,
uint32_t bytes,
uint16_t oflow_bytes,
uint32_t pkts,
uint16_t oflow_pkts,
uint64_t time,
bool teardown_time,
const RevFlowDepParams *p) {
FlowEntry *flow = info->flow();
FlowEntry *rflow = info->reverse_flow();
FLOW_LOCK(flow, rflow, FlowEvent::FLOW_MESSAGE);
UpdateAndExportInternal(info, bytes, oflow_bytes, pkts, oflow_pkts, time,
teardown_time, p);
}
void FlowStatsCollector::UpdateAndExportInternal(FlowExportInfo *info,
uint32_t bytes,
uint16_t oflow_bytes,
uint32_t pkts,
uint16_t oflow_pkts,
uint64_t time,
bool teardown_time,
const RevFlowDepParams *p) {
uint64_t diff_bytes, diff_pkts;
UpdateFlowStatsInternal(info, bytes, oflow_bytes, pkts, oflow_pkts, time,
teardown_time, &diff_bytes, &diff_pkts);
ExportFlow(info, diff_bytes, diff_pkts, p);
}
// Scan for max_count entries in flow-table
uint32_t FlowStatsCollector::RunAgeing(uint32_t max_count) {
FlowEntryTree::iterator it = flow_tree_.lower_bound(flow_iteration_key_);
if (it == flow_tree_.end()) {
it = flow_tree_.begin();
}
KSyncFlowMemory *ksync_obj = agent_uve_->agent()->ksync()->
ksync_flow_memory();
uint32_t count = 0;
uint64_t curr_time = GetCurrentTime();
while (it != flow_tree_.end() && count < max_count) {
FlowExportInfo *info = NULL;
info = &it->second;
FlowEntry *fe = info->flow();
FlowEntry *rfe = info->reverse_flow();
uint32_t flow_handle;
uint16_t gen_id;
{
FlowEntry *rflow = NULL;
FLOW_LOCK(fe, rflow, FlowEvent::FLOW_MESSAGE);
// since flow processing and stats collector can run in parallel
// flow handle and gen id not being the key for flow entry can
// change while processing, so flow handle and gen id should be
// fetched by holding an lock and should not be re-fetched again
// during the entry processing
flow_handle = fe->flow_handle();
gen_id = fe->gen_id();
}
it++;
// if we come across deleted entry, retry flow deletion after some time
// duplicate delete will be suppressed in flow_table
uint64_t delete_time = info->delete_enqueue_time();
if (delete_time) {
if ((curr_time - delete_time) > kFlowDeleteRetryTime) {
FlowDeleteEnqueue(info, curr_time);
count++;
}
continue;
}
count++;
const vr_flow_entry *k_flow = ksync_obj->GetValidKFlowEntry
(fe->key(), flow_handle, gen_id);
if ((fe->key().protocol == IPPROTO_TCP) &&
ksync_obj->IsEvictionMarked(k_flow)) {
uint64_t evict_time = info->evict_enqueue_time();
if (evict_time) {
if ((curr_time - evict_time) > kFlowDeleteRetryTime) {
FlowEvictEnqueue(info, curr_time, flow_handle, gen_id);
}
continue;
}
FlowEvictEnqueue(info, curr_time, flow_handle, gen_id);
continue;
}
FlowExportInfo *rev_info = NULL;
// Delete short flows
if ((flow_stats_manager_->delete_short_flow() == true) &&
fe->is_flags_set(FlowEntry::ShortFlow)) {
rev_info = FindFlowExportInfo(rfe);
FlowDeleteEnqueue(info, curr_time);
if (rev_info) {
count++;
}
continue;
}
bool deleted = false;
// Can the flow be aged?
if (ShouldBeAged(info, k_flow, curr_time)) {
rev_info = FindFlowExportInfo(rfe);
// ShouldBeAged looks at one flow only. So, check for both forward and
// reverse flows
if (rev_info) {
const vr_flow_entry *k_flow_rev;
k_flow_rev = ksync_obj->GetValidKFlowEntry
(rfe->key(), rfe->flow_handle(), rfe->gen_id());
if (ShouldBeAged(rev_info, k_flow_rev, curr_time)) {
deleted = true;
}
} else {
deleted = true;
}
}
if (deleted == true) {
FlowDeleteEnqueue(info, curr_time);
// We delete both forward and reverse flows. So, account for
// reverse flow also
if (rev_info) {
count++;
}
}
// Update stats for flows not being deleted
// Stats for deleted flow are updated when we get DELETE message
if (deleted == false && k_flow) {
uint64_t k_bytes, bytes;
/* Copy full stats in one shot and use local copy instead of reading
* individual stats from shared memory directly to minimize the
* inconsistency */
struct vr_flow_stats fe_stats = k_flow->fe_stats;
k_bytes = GetFlowStats(fe_stats.flow_bytes_oflow,
fe_stats.flow_bytes);
bytes = 0x0000ffffffffffffULL & info->bytes();
/* Always copy udp source port even though vrouter does not change
* it. Vrouter many change this behavior and recompute source port
* whenever flow action changes. To keep agent independent of this,
* always copy UDP source port */
info->set_underlay_source_port(k_flow->fe_udp_src_port);
info->set_tcp_flags(k_flow->fe_tcp_flags);
/* Don't account for agent overflow bits while comparing change in
* stats */
if (bytes != k_bytes) {
UpdateAndExportInternalLocked(info,
fe_stats.flow_bytes,
fe_stats.flow_bytes_oflow,
fe_stats.flow_packets,
fe_stats.flow_packets_oflow,
curr_time, false, NULL);
} else if (info->changed()) {
/* export flow (reverse) for which traffic is not seen yet. */
ExportFlowLocked(info, 0, 0, NULL);
}
}
}
//Send any pending flow export messages
DispatchPendingFlowMsg();
// Update iterator for next pass
if (it == flow_tree_.end()) {
flow_iteration_key_ = NULL;
} else {
flow_iteration_key_ = it->first;
}
return count;
}
// Timer fired for ageing. Update the number of entries to visit and start the
// task if its already not ruuning
bool FlowStatsCollector::Run() {
run_counter_++;
if (flow_tree_.size() == 0) {
return true;
}
// Update number of entries to visit in flow.
UpdateEntriesToVisit();
// Start task to scan the entries
if (ageing_task_ == NULL) {
ageing_task_ = new AgeingTask(this);
agent_uve_->agent()->task_scheduler()->Enqueue(ageing_task_);
}
return true;
}
// Called on runnig of a task
bool FlowStatsCollector::RunAgeingTask() {
// Run ageing per task
uint32_t count = RunAgeing(kFlowsPerTask);
// Update number of entries visited
if (count < entries_to_visit_)
entries_to_visit_ -= count;
else
entries_to_visit_ = 0;
// Done with task if we reach end of tree or count is exceeded
if (flow_iteration_key_ == NULL || entries_to_visit_ == 0) {
entries_to_visit_ = 0;
ageing_task_ = NULL;
return true;
}
// More entries to visit. Continue the task
return false;
}
/////////////////////////////////////////////////////////////////////////////
// Utility methods to enqueue events into work-queue
/////////////////////////////////////////////////////////////////////////////
void FlowStatsCollector::AddEvent(const FlowEntryPtr &flow) {
FlowExportInfo info(flow, GetCurrentTime());
boost::shared_ptr<FlowExportReq>
req(new FlowExportReq(FlowExportReq::ADD_FLOW, info));
request_queue_.Enqueue(req);
}
void FlowStatsCollector::DeleteEvent(const FlowEntryPtr &flow,
const RevFlowDepParams ¶ms) {
FlowExportInfo info(flow);
boost::shared_ptr<FlowExportReq>
req(new FlowExportReq(FlowExportReq::DELETE_FLOW, info,
GetCurrentTime(), params));
request_queue_.Enqueue(req);
}
void FlowStatsCollector::UpdateStatsEvent(const FlowEntryPtr &flow,
uint32_t bytes,
uint32_t packets,
uint32_t oflow_bytes) {
FlowExportInfo info(flow);
boost::shared_ptr<FlowExportReq>
req(new FlowExportReq(FlowExportReq::UPDATE_FLOW_STATS, info, bytes,
packets, oflow_bytes));
request_queue_.Enqueue(req);
}
void FlowStatsCollector::SetUnderlayInfo(FlowExportInfo *info,
FlowLogData &s_flow) {
string rid = agent_uve_->agent()->router_id().to_string();
FlowEntry *flow = info->flow();
uint16_t underlay_src_port = 0;
if (flow->is_flags_set(FlowEntry::LocalFlow)) {
s_flow.set_vrouter_ip(rid);
s_flow.set_other_vrouter_ip(rid);
/* Set source_port as 0 for local flows. Source port is calculated by
* vrouter irrespective of whether flow is local or not. So for local
* flows we need to ignore port given by vrouter
*/
s_flow.set_underlay_source_port(0);
} else {
s_flow.set_vrouter_ip(rid);
s_flow.set_other_vrouter_ip(flow->peer_vrouter());
if (flow->tunnel_type().GetType() != TunnelType::MPLS_GRE) {
underlay_src_port = info->underlay_source_port();
}
s_flow.set_underlay_source_port(underlay_src_port);
}
s_flow.set_underlay_proto(flow->tunnel_type().GetType());
}
/* For ingress flows, change the SIP as Nat-IP instead of Native IP */
void FlowStatsCollector::SourceIpOverride(FlowExportInfo *info,
FlowLogData &s_flow,
const RevFlowDepParams *params) {
if (params && params->sip_.is_v4()) {
s_flow.set_sourceip(params->sip_);
return;
}
FlowEntry *flow = info->flow();
FlowEntry *rflow = info->reverse_flow();
if (flow->is_flags_set(FlowEntry::NatFlow) && s_flow.get_direction_ing() &&
rflow) {
const FlowKey *nat_key = &rflow->key();
if (flow->key().src_addr != nat_key->dst_addr) {
s_flow.set_sourceip(nat_key->dst_addr);
}
}
}
void FlowStatsCollector::SetImplicitFlowDetails(FlowExportInfo *info,
FlowLogData &s_flow,
const RevFlowDepParams *params) {
FlowEntry *rflow = info->reverse_flow();
if (rflow) {
s_flow.set_flowuuid(to_string(rflow->egress_uuid()));
s_flow.set_vm(rflow->data().vm_cfg_name);
s_flow.set_sg_rule_uuid(rflow->sg_rule_uuid());
if (rflow->intf_entry()) {
s_flow.set_vmi_uuid(UuidToString(rflow->intf_entry()->GetUuid()));
}
s_flow.set_reverse_uuid(to_string(rflow->uuid()));
} else if (params) {
s_flow.set_flowuuid(to_string(params->rev_egress_uuid_));
s_flow.set_vm(params->vm_cfg_name_);
s_flow.set_sg_rule_uuid(params->sg_uuid_);
s_flow.set_reverse_uuid(to_string(params->rev_uuid_));
s_flow.set_vmi_uuid(params->vmi_uuid_);
}
}
void FlowStatsCollector::GetFlowSandeshActionParams
(const FlowAction &action_info, std::string &action_str) {
std::bitset<32> bs(action_info.action);
for (unsigned int i = 0; i <= bs.size(); i++) {
if (bs[i]) {
if (!action_str.empty()) {
action_str += "|";
}
action_str += TrafficAction::ActionToString(
static_cast<TrafficAction::Action>(i));
}
}
}
void FlowStatsCollector::EnqueueFlowMsg() {
msg_index_++;
if (msg_index_ == kMaxFlowMsgsPerSend) {
DispatchFlowMsg(msg_list_);
msg_index_ = 0;
}
}
void FlowStatsCollector::DispatchPendingFlowMsg() {
if (msg_index_ == 0) {
return;
}
vector<FlowLogData>::const_iterator first = msg_list_.begin();
vector<FlowLogData>::const_iterator last = msg_list_.begin() + msg_index_;
vector<FlowLogData> new_list(first, last);
DispatchFlowMsg(new_list);
msg_index_ = 0;
}
void FlowStatsCollector::DispatchFlowMsg(const std::vector<FlowLogData> &lst) {
FLOW_LOG_DATA_OBJECT_LOG("", SandeshLevel::SYS_INFO, lst);
}
uint8_t FlowStatsCollector::GetFlowMsgIdx() {
FlowLogData &obj = msg_list_[msg_index_];
obj = FlowLogData();
return msg_index_;
}
void FlowStatsCollector::ExportFlowLocked(FlowExportInfo *info,
uint64_t diff_bytes,
uint64_t diff_pkts,
const RevFlowDepParams *params) {
FlowEntry *flow = info->flow();
FlowEntry *rflow = info->reverse_flow();
FLOW_LOCK(flow, rflow, FlowEvent::FLOW_MESSAGE);
ExportFlow(info, diff_bytes, diff_pkts, params);
}
/* Flow Export Algorithm
* (1) Flow samples greater than or equal to sampling threshold will always be
* exported, with the byte/packet counts reported as-is.
* (2) Flow samples smaller than the sampling threshold will be exported
* probabilistically, with the byte/packets counts adjusted upwards according to
* the probability.
* (3) Probability = diff_bytes/sampling_threshold
* (4) We generate a random number less than sampling threshold.
* (5) If the diff_bytes is greater than random number then the flow is dropped
* (6) Otherwise the flow is exported after normalizing the diff bytes and
* packets. The normalization is done by dividing diff_bytes and diff_pkts with
* probability. This normalization is used as heuristictic to account for stats
* of dropped flows */
void FlowStatsCollector::ExportFlow(FlowExportInfo *info,
uint64_t diff_bytes,
uint64_t diff_pkts,
const RevFlowDepParams *params) {
assert((agent_uve_->agent()->tsn_enabled() == false));
FlowEntry *flow = info->flow();
FlowEntry *rflow = info->reverse_flow();
int32_t cfg_rate = agent_uve_->agent()->oper_db()->global_vrouter()->
flow_export_rate();
/* We should always try to export flows with Action as LOG regardless of
* configured flow-export-rate */
if (!info->IsActionLog() && !cfg_rate) {
flow_stats_manager_->flow_export_disable_drops_++;
return;
}
/* Subject a flow to sampling algorithm only when all of below is met:-
* a. if Log is not configured as action for flow
* b. actual flow-export-rate is >= 80% of configured flow-export-rate. This
* is done only for first time.
* c. diff_bytes is lesser than the threshold
* d. Flow-sampling is not disabled
* e. Flow-sample does not have teardown time or the sample for the flow is
* not exported earlier.
*/
bool subject_flows_to_algorithm = false;
if (!info->IsActionLog() && (diff_bytes < threshold()) &&
(cfg_rate != GlobalVrouter::kDisableSampling) &&
(!info->teardown_time() || !info->exported_atleast_once()) &&
((!flow_stats_manager_->flows_sampled_atleast_once_ &&
flow_stats_manager_->flow_export_rate() >= ((double)cfg_rate) * 0.8)
|| flow_stats_manager_->flows_sampled_atleast_once_)) {
subject_flows_to_algorithm = true;
flow_stats_manager_->set_flows_sampled_atleast_once();
}
if (subject_flows_to_algorithm) {
double probability = diff_bytes/threshold();
uint32_t num = rand() % threshold();
if (num > diff_bytes) {
/* Do not export the flow, if the random number generated is more
* than the diff_bytes */
flow_stats_manager_->flow_export_sampling_drops_++;
/* The second part of the if condition below is not required but
* added for better readability. It is not required because
* exported_atleast_once() will always be false if teardown time is
* set. If both teardown_time and exported_atleast_once are true we
* will never be here */
if (info->teardown_time() && !info->exported_atleast_once()) {
flow_stats_manager_->flow_export_drops_++;
}
return;
}
/* Normalize the diff_bytes and diff_packets reported using the
* probability value */
if (probability == 0) {
diff_bytes = diff_pkts = 0;
} else {
diff_bytes = diff_bytes/probability;
diff_pkts = diff_pkts/probability;
}
}
/* Mark the flow as exported */
info->set_exported_atleast_once(true);
FlowLogData &s_flow = msg_list_[GetFlowMsgIdx()];
s_flow.set_flowuuid(to_string(flow->uuid()));
s_flow.set_bytes(info->bytes());
s_flow.set_packets(info->packets());
s_flow.set_diff_bytes(diff_bytes);
s_flow.set_diff_packets(diff_pkts);
s_flow.set_tcp_flags(info->tcp_flags());
s_flow.set_sourceip(flow->key().src_addr);
s_flow.set_destip(flow->key().dst_addr);
s_flow.set_protocol(flow->key().protocol);
s_flow.set_sport(flow->key().src_port);
s_flow.set_dport(flow->key().dst_port);
s_flow.set_sourcevn(flow->data().source_vn_match);
s_flow.set_destvn(flow->data().dest_vn_match);
s_flow.set_vm(flow->data().vm_cfg_name);
if (flow->is_flags_set(FlowEntry::ReverseFlow)) {
s_flow.set_forward_flow(false);
} else {
s_flow.set_forward_flow(true);
}
string drop_reason = FlowEntry::DropReasonStr(flow->data().drop_reason);
s_flow.set_drop_reason(drop_reason);
s_flow.set_sg_rule_uuid(flow->sg_rule_uuid());
s_flow.set_nw_ace_uuid(flow->nw_ace_uuid());
if (flow->intf_entry()) {
s_flow.set_vmi_uuid(UuidToString(flow->intf_entry()->GetUuid()));
}
if (rflow) {
s_flow.set_reverse_uuid(to_string(rflow->uuid()));
} else if (params) {
s_flow.set_reverse_uuid(to_string(params->rev_uuid_));
}
// Set flow action
std::string action_str;
GetFlowSandeshActionParams(flow->data().match_p.action_info, action_str);
s_flow.set_action(action_str);
s_flow.set_setup_time(info->setup_time());
if (info->teardown_time()) {
s_flow.set_teardown_time(info->teardown_time());
}
SetUnderlayInfo(info, s_flow);
info->set_changed(false);
if (flow->is_flags_set(FlowEntry::LocalFlow)) {
/* For local flows we need to send two flow log messages.
* 1. With direction as ingress
* 2. With direction as egress
* For local flows we have already sent flow log above with
* direction as ingress. We are sending flow log below with
* direction as egress.
*/
s_flow.set_direction_ing(1);
s_flow.set_reverse_uuid(to_string(flow->egress_uuid()));
SourceIpOverride(info, s_flow, params);
EnqueueFlowMsg();
FlowLogData &s_flow2 = msg_list_[GetFlowMsgIdx()];
s_flow2 = s_flow;
s_flow2.set_direction_ing(0);
//Update the interface and VM name in this flow
//For the reverse flow this would be egress flow
//For example
// VM1 A --> B is fwd flow
// VM2 B --> A is rev flow
//
//Egress flow for fwd flow would be exported
//while exporting rev flow, this done so that
//key, stats and other stuff can be copied over
//from current flow
SetImplicitFlowDetails(info, s_flow2, params);
//Export local flow of egress direction with a different UUID even when
//the flow is same. Required for analytics module to query flows
//irrespective of direction.
EnqueueFlowMsg();
flow_stats_manager_->UpdateFlowExportStats(2,
subject_flows_to_algorithm);
} else {
if (flow->is_flags_set(FlowEntry::IngressDir)) {
s_flow.set_direction_ing(1);
SourceIpOverride(info, s_flow, params);
} else {
s_flow.set_direction_ing(0);
}
EnqueueFlowMsg();
flow_stats_manager_->UpdateFlowExportStats(1,
subject_flows_to_algorithm);
}
}
bool FlowStatsManager::UpdateFlowThreshold() {
uint64_t curr_time = FlowStatsCollector::GetCurrentTime();
bool export_rate_calculated = false;
uint32_t exp_rate_without_sampling = 0;
/* If flows are not being exported, no need to update threshold */
if (!flow_export_count_) {
return true;
}
// Calculate Flow Export rate
if (prev_flow_export_rate_compute_time_) {
uint64_t diff_secs = 0;
uint64_t diff_micro_secs = curr_time -
prev_flow_export_rate_compute_time_;
if (diff_micro_secs) {
diff_secs = diff_micro_secs/1000000;
}
if (diff_secs) {
uint32_t flow_export_count = flow_export_count_reset();
flow_export_rate_ = flow_export_count/diff_secs;
exp_rate_without_sampling =
flow_export_without_sampling_reset()/diff_secs;
prev_flow_export_rate_compute_time_ = curr_time;
export_rate_calculated = true;
}
} else {
prev_flow_export_rate_compute_time_ = curr_time;
flow_export_count_ = 0;
return true;
}
uint32_t cfg_rate = agent_->oper_db()->global_vrouter()->
flow_export_rate();
/* No need to update threshold when flow_export_rate is NOT calculated
* and configured flow export rate has not changed */
if (!export_rate_calculated &&
(cfg_rate == prev_cfg_flow_export_rate_)) {
return true;
}
uint64_t cur_t = threshold(), new_t = 0;
// Update sampling threshold based on flow_export_rate_
if (flow_export_rate_ < ((double)cfg_rate) * 0.8) {
/* There are two reasons why we can be here.
* 1. None of the flows were sampled because we never crossed
* 80% of configured flow-export-rate.
* 2. In scale setups, the threshold was updated to high value because
* of which flow-export-rate has dropped drastically.
* Threshold should be updated here depending on which of the above two
* situations we are in. */
if (!flows_sampled_atleast_once_) {
UpdateThreshold(kDefaultFlowSamplingThreshold, false);
} else {
if (flow_export_rate_ < ((double)cfg_rate) * 0.5) {
UpdateThreshold((threshold_ / 4), false);
} else {
UpdateThreshold((threshold_ / 2), false);
}
}
} else if (flow_export_rate_ > (cfg_rate * 3)) {
UpdateThreshold((threshold_ * 4), true);
} else if (flow_export_rate_ > (cfg_rate * 2)) {
UpdateThreshold((threshold_ * 3), true);
} else if (flow_export_rate_ > ((double)cfg_rate) * 1.25) {
UpdateThreshold((threshold_ * 2), true);
}
prev_cfg_flow_export_rate_ = cfg_rate;
new_t = threshold();
FLOW_EXPORT_STATS_TRACE(flow_export_rate_, exp_rate_without_sampling, cur_t,
new_t);
return true;
}
uint64_t FlowStatsCollector::threshold() const {
return flow_stats_manager_->threshold();
}
uint32_t FlowStatsCollector::flow_export_count() const {
return flow_stats_manager_->flow_export_count();
}
bool FlowStatsCollector::RequestHandler(boost::shared_ptr<FlowExportReq> req) {
const FlowExportInfo &info = req->info();
FlowEntry *flow = info.flow();
FlowEntry *rflow = info.reverse_flow();