summaryrefslogtreecommitdiffstats
path: root/src/rgw/rgw_quota.cc
blob: 052bc7a5be65e585ea0e29ed3354cbd47be44cb2 (plain)
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
// vim: ts=8 sw=2 smarttab

/*
 * Ceph - scalable distributed file system
 *
 * Copyright (C) 2013 Inktank, Inc
 *
 * This is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License version 2.1, as published by the Free Software
 * Foundation.  See file COPYING.
 *
 */


#include "include/utime.h"
#include "common/lru_map.h"
#include "common/RefCountedObj.h"
#include "common/Thread.h"
#include "common/Mutex.h"
#include "common/RWLock.h"

#include "rgw_common.h"
#include "rgw_rados.h"
#include "rgw_quota.h"
#include "rgw_bucket.h"
#include "rgw_user.h"

#include "services/svc_sys_obj.h"

#include <atomic>

#define dout_context g_ceph_context
#define dout_subsys ceph_subsys_rgw


struct RGWQuotaCacheStats {
  RGWStorageStats stats;
  utime_t expiration;
  utime_t async_refresh_time;
};

template<class T>
class RGWQuotaCache {
protected:
  RGWRados *store;
  lru_map<T, RGWQuotaCacheStats> stats_map;
  RefCountedWaitObject *async_refcount;

  class StatsAsyncTestSet : public lru_map<T, RGWQuotaCacheStats>::UpdateContext {
    int objs_delta;
    uint64_t added_bytes;
    uint64_t removed_bytes;
  public:
    StatsAsyncTestSet() : objs_delta(0), added_bytes(0), removed_bytes(0) {}
    bool update(RGWQuotaCacheStats *entry) override {
      if (entry->async_refresh_time.sec() == 0)
        return false;

      entry->async_refresh_time = utime_t(0, 0);

      return true;
    }
  };

  virtual int fetch_stats_from_storage(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats) = 0;

  virtual bool map_find(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs) = 0;

  virtual bool map_find_and_update(const rgw_user& user, const rgw_bucket& bucket, typename lru_map<T, RGWQuotaCacheStats>::UpdateContext *ctx) = 0;
  virtual void map_add(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs) = 0;

  virtual void data_modified(const rgw_user& user, rgw_bucket& bucket) {}
public:
  RGWQuotaCache(RGWRados *_store, int size) : store(_store), stats_map(size) {
    async_refcount = new RefCountedWaitObject;
  }
  virtual ~RGWQuotaCache() {
    async_refcount->put_wait(); /* wait for all pending async requests to complete */
  }

  int get_stats(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats, RGWQuotaInfo& quota);
  void adjust_stats(const rgw_user& user, rgw_bucket& bucket, int objs_delta, uint64_t added_bytes, uint64_t removed_bytes);

  virtual bool can_use_cached_stats(RGWQuotaInfo& quota, RGWStorageStats& stats);

  void set_stats(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs, RGWStorageStats& stats);
  int async_refresh(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs);
  void async_refresh_response(const rgw_user& user, rgw_bucket& bucket, RGWStorageStats& stats);
  void async_refresh_fail(const rgw_user& user, rgw_bucket& bucket);

  class AsyncRefreshHandler {
  protected:
    RGWRados *store;
    RGWQuotaCache<T> *cache;
  public:
    AsyncRefreshHandler(RGWRados *_store, RGWQuotaCache<T> *_cache) : store(_store), cache(_cache) {}
    virtual ~AsyncRefreshHandler() {}

    virtual int init_fetch() = 0;
    virtual void drop_reference() = 0;
  };

  virtual AsyncRefreshHandler *allocate_refresh_handler(const rgw_user& user, const rgw_bucket& bucket) = 0;
};

template<class T>
bool RGWQuotaCache<T>::can_use_cached_stats(RGWQuotaInfo& quota, RGWStorageStats& cached_stats)
{
  if (quota.max_size >= 0) {
    if (quota.max_size_soft_threshold < 0) {
      quota.max_size_soft_threshold = quota.max_size * store->ctx()->_conf->rgw_bucket_quota_soft_threshold;
    }

    if (cached_stats.size_rounded  >= (uint64_t)quota.max_size_soft_threshold) {
      ldout(store->ctx(), 20) << "quota: can't use cached stats, exceeded soft threshold (size): "
        << cached_stats.size_rounded << " >= " << quota.max_size_soft_threshold << dendl;
      return false;
    }
  }

  if (quota.max_objects >= 0) {
    if (quota.max_objs_soft_threshold < 0) {
      quota.max_objs_soft_threshold = quota.max_objects * store->ctx()->_conf->rgw_bucket_quota_soft_threshold;
    }

    if (cached_stats.num_objects >= (uint64_t)quota.max_objs_soft_threshold) {
      ldout(store->ctx(), 20) << "quota: can't use cached stats, exceeded soft threshold (num objs): "
        << cached_stats.num_objects << " >= " << quota.max_objs_soft_threshold << dendl;
      return false;
    }
  }

  return true;
}

template<class T>
int RGWQuotaCache<T>::async_refresh(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs)
{
  /* protect against multiple updates */
  StatsAsyncTestSet test_update;
  if (!map_find_and_update(user, bucket, &test_update)) {
    /* most likely we just raced with another update */
    return 0;
  }

  async_refcount->get();


  AsyncRefreshHandler *handler = allocate_refresh_handler(user, bucket);

  int ret = handler->init_fetch();
  if (ret < 0) {
    async_refcount->put();
    handler->drop_reference();
    return ret;
  }

  return 0;
}

template<class T>
void RGWQuotaCache<T>::async_refresh_fail(const rgw_user& user, rgw_bucket& bucket)
{
  ldout(store->ctx(), 20) << "async stats refresh response for bucket=" << bucket << dendl;

  async_refcount->put();
}

template<class T>
void RGWQuotaCache<T>::async_refresh_response(const rgw_user& user, rgw_bucket& bucket, RGWStorageStats& stats)
{
  ldout(store->ctx(), 20) << "async stats refresh response for bucket=" << bucket << dendl;

  RGWQuotaCacheStats qs;

  map_find(user, bucket, qs);

  set_stats(user, bucket, qs, stats);

  async_refcount->put();
}

template<class T>
void RGWQuotaCache<T>::set_stats(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs, RGWStorageStats& stats)
{
  qs.stats = stats;
  qs.expiration = ceph_clock_now();
  qs.async_refresh_time = qs.expiration;
  qs.expiration += store->ctx()->_conf->rgw_bucket_quota_ttl;
  qs.async_refresh_time += store->ctx()->_conf->rgw_bucket_quota_ttl / 2;

  map_add(user, bucket, qs);
}

template<class T>
int RGWQuotaCache<T>::get_stats(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats, RGWQuotaInfo& quota) {
  RGWQuotaCacheStats qs;
  utime_t now = ceph_clock_now();
  if (map_find(user, bucket, qs)) {
    if (qs.async_refresh_time.sec() > 0 && now >= qs.async_refresh_time) {
      int r = async_refresh(user, bucket, qs);
      if (r < 0) {
        ldout(store->ctx(), 0) << "ERROR: quota async refresh returned ret=" << r << dendl;

        /* continue processing, might be a transient error, async refresh is just optimization */
      }
    }

    if (can_use_cached_stats(quota, qs.stats) && qs.expiration >
	ceph_clock_now()) {
      stats = qs.stats;
      return 0;
    }
  }

  int ret = fetch_stats_from_storage(user, bucket, stats);
  if (ret < 0 && ret != -ENOENT)
    return ret;

  set_stats(user, bucket, qs, stats);

  return 0;
}


template<class T>
class RGWQuotaStatsUpdate : public lru_map<T, RGWQuotaCacheStats>::UpdateContext {
  const int objs_delta;
  const uint64_t added_bytes;
  const uint64_t removed_bytes;
public:
  RGWQuotaStatsUpdate(const int objs_delta,
                      const uint64_t added_bytes,
                      const uint64_t removed_bytes)
    : objs_delta(objs_delta),
      added_bytes(added_bytes),
      removed_bytes(removed_bytes) {
  }

  bool update(RGWQuotaCacheStats * const entry) override {
    const uint64_t rounded_added = rgw_rounded_objsize(added_bytes);
    const uint64_t rounded_removed = rgw_rounded_objsize(removed_bytes);

    if (((int64_t)(entry->stats.size + added_bytes - removed_bytes)) >= 0) {
      entry->stats.size += added_bytes - removed_bytes;
    } else {
      entry->stats.size = 0;
    }

    if (((int64_t)(entry->stats.size_rounded + rounded_added - rounded_removed)) >= 0) {
      entry->stats.size_rounded += rounded_added - rounded_removed;
    } else {
      entry->stats.size_rounded = 0;
    }

    if (((int64_t)(entry->stats.num_objects + objs_delta)) >= 0) {
      entry->stats.num_objects += objs_delta;
    } else {
      entry->stats.num_objects = 0;
    }

    return true;
  }
};


template<class T>
void RGWQuotaCache<T>::adjust_stats(const rgw_user& user, rgw_bucket& bucket, int objs_delta,
                                 uint64_t added_bytes, uint64_t removed_bytes)
{
  RGWQuotaStatsUpdate<T> update(objs_delta, added_bytes, removed_bytes);
  map_find_and_update(user, bucket, &update);

  data_modified(user, bucket);
}

class BucketAsyncRefreshHandler : public RGWQuotaCache<rgw_bucket>::AsyncRefreshHandler,
                                  public RGWGetBucketStats_CB {
  rgw_user user;
public:
  BucketAsyncRefreshHandler(RGWRados *_store, RGWQuotaCache<rgw_bucket> *_cache,
                            const rgw_user& _user, const rgw_bucket& _bucket) :
                                      RGWQuotaCache<rgw_bucket>::AsyncRefreshHandler(_store, _cache),
                                      RGWGetBucketStats_CB(_bucket), user(_user) {}

  void drop_reference() override { put(); }
  void handle_response(int r) override;
  int init_fetch() override;
};

int BucketAsyncRefreshHandler::init_fetch()
{
  RGWBucketInfo bucket_info;

  auto obj_ctx = store->svc.sysobj->init_obj_ctx();

  int r = store->get_bucket_instance_info(obj_ctx, bucket, bucket_info, NULL, NULL);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket info for bucket=" << bucket << " r=" << r << dendl;
    return r;
  }

  ldout(store->ctx(), 20) << "initiating async quota refresh for bucket=" << bucket << dendl;

  r = store->get_bucket_stats_async(bucket_info, RGW_NO_SHARD, this);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket info for bucket=" << bucket.name << dendl;

    /* get_bucket_stats_async() dropped our reference already */
    return r;
  }

  return 0;
}

void BucketAsyncRefreshHandler::handle_response(const int r)
{
  if (r < 0) {
    ldout(store->ctx(), 20) << "AsyncRefreshHandler::handle_response() r=" << r << dendl;
    cache->async_refresh_fail(user, bucket);
    return;
  }

  RGWStorageStats bs;

  for (const auto& pair : *stats) {
    const RGWStorageStats& s = pair.second;

    bs.size += s.size;
    bs.size_rounded += s.size_rounded;
    bs.num_objects += s.num_objects;
  }

  cache->async_refresh_response(user, bucket, bs);
}

class RGWBucketStatsCache : public RGWQuotaCache<rgw_bucket> {
protected:
  bool map_find(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs) override {
    return stats_map.find(bucket, qs);
  }

  bool map_find_and_update(const rgw_user& user, const rgw_bucket& bucket, lru_map<rgw_bucket, RGWQuotaCacheStats>::UpdateContext *ctx) override {
    return stats_map.find_and_update(bucket, NULL, ctx);
  }

  void map_add(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs) override {
    stats_map.add(bucket, qs);
  }

  int fetch_stats_from_storage(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats) override;

public:
  explicit RGWBucketStatsCache(RGWRados *_store) : RGWQuotaCache<rgw_bucket>(_store, _store->ctx()->_conf->rgw_bucket_quota_cache_size) {
  }

  AsyncRefreshHandler *allocate_refresh_handler(const rgw_user& user, const rgw_bucket& bucket) override {
    return new BucketAsyncRefreshHandler(store, this, user, bucket);
  }
};

int RGWBucketStatsCache::fetch_stats_from_storage(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats)
{
  RGWBucketInfo bucket_info;

  RGWSysObjectCtx obj_ctx = store->svc.sysobj->init_obj_ctx();

  int r = store->get_bucket_instance_info(obj_ctx, bucket, bucket_info, NULL, NULL);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket info for bucket=" << bucket << " r=" << r << dendl;
    return r;
  }

  string bucket_ver;
  string master_ver;

  map<RGWObjCategory, RGWStorageStats> bucket_stats;
  r = store->get_bucket_stats(bucket_info, RGW_NO_SHARD, &bucket_ver,
                                  &master_ver, bucket_stats, nullptr);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket stats for bucket="
                           << bucket.name << dendl;
    return r;
  }

  stats = RGWStorageStats();

  for (const auto& pair : bucket_stats) {
    const RGWStorageStats& s = pair.second;

    stats.size += s.size;
    stats.size_rounded += s.size_rounded;
    stats.num_objects += s.num_objects;
  }

  return 0;
}

class UserAsyncRefreshHandler : public RGWQuotaCache<rgw_user>::AsyncRefreshHandler,
                                public RGWGetUserStats_CB {
  rgw_bucket bucket;
public:
  UserAsyncRefreshHandler(RGWRados *_store, RGWQuotaCache<rgw_user> *_cache,
                          const rgw_user& _user, const rgw_bucket& _bucket) :
                          RGWQuotaCache<rgw_user>::AsyncRefreshHandler(_store, _cache),
                          RGWGetUserStats_CB(_user),
                          bucket(_bucket) {}

  void drop_reference() override { put(); }
  int init_fetch() override;
  void handle_response(int r) override;
};

int UserAsyncRefreshHandler::init_fetch()
{
  ldout(store->ctx(), 20) << "initiating async quota refresh for user=" << user << dendl;
  int r = store->get_user_stats_async(user, this);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket info for user=" << user << dendl;

    /* get_bucket_stats_async() dropped our reference already */
    return r;
  }

  return 0;
}

void UserAsyncRefreshHandler::handle_response(int r)
{
  if (r < 0) {
    ldout(store->ctx(), 20) << "AsyncRefreshHandler::handle_response() r=" << r << dendl;
    cache->async_refresh_fail(user, bucket);
    return;
  }

  cache->async_refresh_response(user, bucket, stats);
}

class RGWUserStatsCache : public RGWQuotaCache<rgw_user> {
  std::atomic<bool> down_flag = { false };
  RWLock rwlock;
  map<rgw_bucket, rgw_user> modified_buckets;

  /* thread, sync recent modified buckets info */
  class BucketsSyncThread : public Thread {
    CephContext *cct;
    RGWUserStatsCache *stats;

    Mutex lock;
    Cond cond;
  public:

    BucketsSyncThread(CephContext *_cct, RGWUserStatsCache *_s) : cct(_cct), stats(_s), lock("RGWUserStatsCache::BucketsSyncThread") {}

    void *entry() override {
      ldout(cct, 20) << "BucketsSyncThread: start" << dendl;
      do {
        map<rgw_bucket, rgw_user> buckets;

        stats->swap_modified_buckets(buckets);

        for (map<rgw_bucket, rgw_user>::iterator iter = buckets.begin(); iter != buckets.end(); ++iter) {
          rgw_bucket bucket = iter->first;
          rgw_user& user = iter->second;
          ldout(cct, 20) << "BucketsSyncThread: sync user=" << user << " bucket=" << bucket << dendl;
          int r = stats->sync_bucket(user, bucket);
          if (r < 0) {
            ldout(cct, 0) << "WARNING: sync_bucket() returned r=" << r << dendl;
          }
        }

        if (stats->going_down())
          break;

        lock.Lock();
        cond.WaitInterval(lock, utime_t(cct->_conf->rgw_user_quota_bucket_sync_interval, 0));
        lock.Unlock();
      } while (!stats->going_down());
      ldout(cct, 20) << "BucketsSyncThread: done" << dendl;

      return NULL;
    }

    void stop() {
      Mutex::Locker l(lock);
      cond.Signal();
    }
  };

  /*
   * thread, full sync all users stats periodically
   *
   * only sync non idle users or ones that never got synced before, this is needed so that
   * users that didn't have quota turned on before (or existed before the user objclass
   * tracked stats) need to get their backend stats up to date.
   */
  class UserSyncThread : public Thread {
    CephContext *cct;
    RGWUserStatsCache *stats;

    Mutex lock;
    Cond cond;
  public:

    UserSyncThread(CephContext *_cct, RGWUserStatsCache *_s) : cct(_cct), stats(_s), lock("RGWUserStatsCache::UserSyncThread") {}

    void *entry() override {
      ldout(cct, 20) << "UserSyncThread: start" << dendl;
      do {
        int ret = stats->sync_all_users();
        if (ret < 0) {
          ldout(cct, 5) << "ERROR: sync_all_users() returned ret=" << ret << dendl;
        }

        if (stats->going_down())
          break;

        lock.Lock();
        cond.WaitInterval(lock, utime_t(cct->_conf->rgw_user_quota_sync_interval, 0));
        lock.Unlock();
      } while (!stats->going_down());
      ldout(cct, 20) << "UserSyncThread: done" << dendl;

      return NULL;
    }

    void stop() {
      Mutex::Locker l(lock);
      cond.Signal();
    }
  };

  BucketsSyncThread *buckets_sync_thread;
  UserSyncThread *user_sync_thread;
protected:
  bool map_find(const rgw_user& user,const rgw_bucket& bucket, RGWQuotaCacheStats& qs) override {
    return stats_map.find(user, qs);
  }

  bool map_find_and_update(const rgw_user& user, const rgw_bucket& bucket, lru_map<rgw_user, RGWQuotaCacheStats>::UpdateContext *ctx) override {
    return stats_map.find_and_update(user, NULL, ctx);
  }

  void map_add(const rgw_user& user, const rgw_bucket& bucket, RGWQuotaCacheStats& qs) override {
    stats_map.add(user, qs);
  }

  int fetch_stats_from_storage(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats) override;
  int sync_bucket(const rgw_user& rgw_user, rgw_bucket& bucket);
  int sync_user(const rgw_user& user);
  int sync_all_users();

  void data_modified(const rgw_user& user, rgw_bucket& bucket) override;

  void swap_modified_buckets(map<rgw_bucket, rgw_user>& out) {
    rwlock.get_write();
    modified_buckets.swap(out);
    rwlock.unlock();
  }

  template<class T> /* easier doing it as a template, Thread doesn't have ->stop() */
  void stop_thread(T **pthr) {
    T *thread = *pthr;
    if (!thread)
      return;

    thread->stop();
    thread->join();
    delete thread;
    *pthr = NULL;
  }

public:
  RGWUserStatsCache(RGWRados *_store, bool quota_threads) : RGWQuotaCache<rgw_user>(_store, _store->ctx()->_conf->rgw_bucket_quota_cache_size),
                                        rwlock("RGWUserStatsCache::rwlock") {
    if (quota_threads) {
      buckets_sync_thread = new BucketsSyncThread(store->ctx(), this);
      buckets_sync_thread->create("rgw_buck_st_syn");
      user_sync_thread = new UserSyncThread(store->ctx(), this);
      user_sync_thread->create("rgw_user_st_syn");
    } else {
      buckets_sync_thread = NULL;
      user_sync_thread = NULL;
    }
  }
  ~RGWUserStatsCache() override {
    stop();
  }

  AsyncRefreshHandler *allocate_refresh_handler(const rgw_user& user, const rgw_bucket& bucket) override {
    return new UserAsyncRefreshHandler(store, this, user, bucket);
  }

  bool can_use_cached_stats(RGWQuotaInfo& quota, RGWStorageStats& stats) override {
    /* in the user case, the cached stats may contain a better estimation of the totals, as
     * the backend is only periodically getting updated.
     */
    return true;
  }

  bool going_down() {
    return down_flag;
  }

  void stop() {
    down_flag = true;
    rwlock.get_write();
    stop_thread(&buckets_sync_thread);
    rwlock.unlock();
    stop_thread(&user_sync_thread);
  }
};

int RGWUserStatsCache::fetch_stats_from_storage(const rgw_user& user, const rgw_bucket& bucket, RGWStorageStats& stats)
{
  int r = store->get_user_stats(user, stats);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get user stats for user=" << user << dendl;
    return r;
  }

  return 0;
}

int RGWUserStatsCache::sync_bucket(const rgw_user& user, rgw_bucket& bucket)
{
  RGWBucketInfo bucket_info;

  RGWSysObjectCtx obj_ctx = store->svc.sysobj->init_obj_ctx();

  int r = store->get_bucket_instance_info(obj_ctx, bucket, bucket_info, NULL, NULL);
  if (r < 0) {
    ldout(store->ctx(), 0) << "could not get bucket info for bucket=" << bucket << " r=" << r << dendl;
    return r;
  }

  r = rgw_bucket_sync_user_stats(store, user, bucket_info);
  if (r < 0) {
    ldout(store->ctx(), 0) << "ERROR: rgw_bucket_sync_user_stats() for user=" << user << ", bucket=" << bucket << " returned " << r << dendl;
    return r;
  }

  return 0;
}

int RGWUserStatsCache::sync_user(const rgw_user& user)
{
  cls_user_header header;
  string user_str = user.to_str();
  int ret = store->cls_user_get_header(user_str, &header);
  if (ret < 0) {
    ldout(store->ctx(), 5) << "ERROR: can't read user header: ret=" << ret << dendl;
    return ret;
  }

  if (!store->ctx()->_conf->rgw_user_quota_sync_idle_users &&
      header.last_stats_update < header.last_stats_sync) {
    ldout(store->ctx(), 20) << "user is idle, not doing a full sync (user=" << user << ")" << dendl;
    return 0;
  }

  real_time when_need_full_sync = header.last_stats_sync;
  when_need_full_sync += make_timespan(store->ctx()->_conf->rgw_user_quota_sync_wait_time);
  
  // check if enough time passed since last full sync
  /* FIXME: missing check? */

  ret = rgw_user_sync_all_stats(store, user);
  if (ret < 0) {
    ldout(store->ctx(), 0) << "ERROR: failed user stats sync, ret=" << ret << dendl;
    return ret;
  }

  return 0;
}

int RGWUserStatsCache::sync_all_users()
{
  string key = "user";
  void *handle;

  int ret = store->meta_mgr->list_keys_init(key, &handle);
  if (ret < 0) {
    ldout(store->ctx(), 10) << "ERROR: can't get key: ret=" << ret << dendl;
    return ret;
  }

  bool truncated;
  int max = 1000;

  do {
    list<string> keys;
    ret = store->meta_mgr->list_keys_next(handle, max, keys, &truncated);
    if (ret < 0) {
      ldout(store->ctx(), 0) << "ERROR: lists_keys_next(): ret=" << ret << dendl;
      goto done;
    }
    for (list<string>::iterator iter = keys.begin();
         iter != keys.end() && !going_down(); 
         ++iter) {
      rgw_user user(*iter);
      ldout(store->ctx(), 20) << "RGWUserStatsCache: sync user=" << user << dendl;
      int ret = sync_user(user);
      if (ret < 0) {
        ldout(store->ctx(), 5) << "ERROR: sync_user() failed, user=" << user << " ret=" << ret << dendl;

        /* continuing to next user */
        continue;
      }
    }
  } while (truncated);

  ret = 0;
done:
  store->meta_mgr->list_keys_complete(handle);
  return ret;
}

void RGWUserStatsCache::data_modified(const rgw_user& user, rgw_bucket& bucket)
{
  /* racy, but it's ok */
  rwlock.get_read();
  bool need_update = modified_buckets.find(bucket) == modified_buckets.end();
  rwlock.unlock();

  if (need_update) {
    rwlock.get_write();
    modified_buckets[bucket] = user;
    rwlock.unlock();
  }
}


class RGWQuotaInfoApplier {
  /* NOTE: no non-static field allowed as instances are supposed to live in
   * the static memory only. */
protected:
  RGWQuotaInfoApplier() = default;

public:
  virtual ~RGWQuotaInfoApplier() {}

  virtual bool is_size_exceeded(const char * const entity,
                                const RGWQuotaInfo& qinfo,
                                const RGWStorageStats& stats,
                                const uint64_t size) const = 0;

  virtual bool is_num_objs_exceeded(const char * const entity,
                                    const RGWQuotaInfo& qinfo,
                                    const RGWStorageStats& stats,
                                    const uint64_t num_objs) const = 0;

  static const RGWQuotaInfoApplier& get_instance(const RGWQuotaInfo& qinfo);
};

class RGWQuotaInfoDefApplier : public RGWQuotaInfoApplier {
public:
  bool is_size_exceeded(const char * const entity,
                                const RGWQuotaInfo& qinfo,
                                const RGWStorageStats& stats,
                                const uint64_t size) const override;

  bool is_num_objs_exceeded(const char * const entity,
                                    const RGWQuotaInfo& qinfo,
                                    const RGWStorageStats& stats,
                                    const uint64_t num_objs) const override;
};

class RGWQuotaInfoRawApplier : public RGWQuotaInfoApplier {
public:
  bool is_size_exceeded(const char * const entity,
                                const RGWQuotaInfo& qinfo,
                                const RGWStorageStats& stats,
                                const uint64_t size) const override;

  bool is_num_objs_exceeded(const char * const entity,
                                    const RGWQuotaInfo& qinfo,
                                    const RGWStorageStats& stats,
                                    const uint64_t num_objs) const override;
};


bool RGWQuotaInfoDefApplier::is_size_exceeded(const char * const entity,
                                              const RGWQuotaInfo& qinfo,
                                              const RGWStorageStats& stats,
                                              const uint64_t size) const
{
  if (qinfo.max_size < 0) {
    /* The limit is not enabled. */
    return false;
  }

  const uint64_t cur_size = stats.size_rounded;
  const uint64_t new_size = rgw_rounded_objsize(size);

  if (cur_size + new_size > static_cast<uint64_t>(qinfo.max_size)) {
    dout(10) << "quota exceeded: stats.size_rounded=" << stats.size_rounded
             << " size=" << new_size << " "
             << entity << "_quota.max_size=" << qinfo.max_size << dendl;
    return true;
  }

  return false;
}

bool RGWQuotaInfoDefApplier::is_num_objs_exceeded(const char * const entity,
                                                  const RGWQuotaInfo& qinfo,
                                                  const RGWStorageStats& stats,
                                                  const uint64_t num_objs) const
{
  if (qinfo.max_objects < 0) {
    /* The limit is not enabled. */
    return false;
  }

  if (stats.num_objects + num_objs > static_cast<uint64_t>(qinfo.max_objects)) {
    dout(10) << "quota exceeded: stats.num_objects=" << stats.num_objects
             << " " << entity << "_quota.max_objects=" << qinfo.max_objects
             << dendl;
    return true;
  }

  return false;
}

bool RGWQuotaInfoRawApplier::is_size_exceeded(const char * const entity,
                                              const RGWQuotaInfo& qinfo,
                                              const RGWStorageStats& stats,
                                              const uint64_t size) const
{
  if (qinfo.max_size < 0) {
    /* The limit is not enabled. */
    return false;
  }

  const uint64_t cur_size = stats.size;

  if (cur_size + size > static_cast<uint64_t>(qinfo.max_size)) {
    dout(10) << "quota exceeded: stats.size=" << stats.size
             << " size=" << size << " "
             << entity << "_quota.max_size=" << qinfo.max_size << dendl;
    return true;
  }

  return false;
}

bool RGWQuotaInfoRawApplier::is_num_objs_exceeded(const char * const entity,
                                                  const RGWQuotaInfo& qinfo,
                                                  const RGWStorageStats& stats,
                                                  const uint64_t num_objs) const
{
  if (qinfo.max_objects < 0) {
    /* The limit is not enabled. */
    return false;
  }

  if (stats.num_objects + num_objs > static_cast<uint64_t>(qinfo.max_objects)) {
    dout(10) << "quota exceeded: stats.num_objects=" << stats.num_objects
             << " " << entity << "_quota.max_objects=" << qinfo.max_objects
             << dendl;
    return true;
  }

  return false;
}

const RGWQuotaInfoApplier& RGWQuotaInfoApplier::get_instance(
  const RGWQuotaInfo& qinfo)
{
  static RGWQuotaInfoDefApplier default_qapplier;
  static RGWQuotaInfoRawApplier raw_qapplier;

  if (qinfo.check_on_raw) {
    return raw_qapplier;
  } else {
    return default_qapplier;
  }
}


class RGWQuotaHandlerImpl : public RGWQuotaHandler {
  RGWRados *store;
  RGWBucketStatsCache bucket_stats_cache;
  RGWUserStatsCache user_stats_cache;

  int check_quota(const char * const entity,
                  const RGWQuotaInfo& quota,
                  const RGWStorageStats& stats,
                  const uint64_t num_objs,
                  const uint64_t size) {
    if (!quota.enabled) {
      return 0;
    }

    const auto& quota_applier = RGWQuotaInfoApplier::get_instance(quota);

    ldout(store->ctx(), 20) << entity
                            << " quota: max_objects=" << quota.max_objects
                            << " max_size=" << quota.max_size << dendl;


    if (quota_applier.is_num_objs_exceeded(entity, quota, stats, num_objs)) {
      return -ERR_QUOTA_EXCEEDED;
    }

    if (quota_applier.is_size_exceeded(entity, quota, stats, size)) {
      return -ERR_QUOTA_EXCEEDED;
    }

    ldout(store->ctx(), 20) << entity << " quota OK:"
                            << " stats.num_objects=" << stats.num_objects
                            << " stats.size=" << stats.size << dendl;
    return 0;
  }
public:
  RGWQuotaHandlerImpl(RGWRados *_store, bool quota_threads) : store(_store),
                                    bucket_stats_cache(_store),
                                    user_stats_cache(_store, quota_threads) {}

  int check_quota(const rgw_user& user,
                          rgw_bucket& bucket,
                          RGWQuotaInfo& user_quota,
                          RGWQuotaInfo& bucket_quota,
                          uint64_t num_objs,
                          uint64_t size) override {

    if (!bucket_quota.enabled && !user_quota.enabled) {
      return 0;
    }

    /*
     * we need to fetch bucket stats if the user quota is enabled, because
     * the whole system relies on us periodically updating the user's bucket
     * stats in the user's header, this happens in get_stats() if we actually
     * fetch that info and not rely on cached data
     */

    if (bucket_quota.enabled) {
      RGWStorageStats bucket_stats;
      int ret = bucket_stats_cache.get_stats(user, bucket, bucket_stats,
                                           bucket_quota);
      if (ret < 0) {
        return ret;
      }
      ret = check_quota("bucket", bucket_quota, bucket_stats, num_objs, size);
      if (ret < 0) {
        return ret;
      }
    }

    if (user_quota.enabled) {
      RGWStorageStats user_stats;
      int ret = user_stats_cache.get_stats(user, bucket, user_stats, user_quota);
      if (ret < 0) {
        return ret;
      }
      ret = check_quota("user", user_quota, user_stats, num_objs, size);
      if (ret < 0) {
        return ret;
      }
    }
    return 0;
  }

  void update_stats(const rgw_user& user, rgw_bucket& bucket, int obj_delta, uint64_t added_bytes, uint64_t removed_bytes) override {
    bucket_stats_cache.adjust_stats(user, bucket, obj_delta, added_bytes, removed_bytes);
    user_stats_cache.adjust_stats(user, bucket, obj_delta, added_bytes, removed_bytes);
  }

  int check_bucket_shards(uint64_t max_objs_per_shard, uint64_t num_shards,
			  const rgw_user& user, const rgw_bucket& bucket, RGWQuotaInfo& bucket_quota,
			  uint64_t num_objs, bool& need_resharding, uint32_t *suggested_num_shards) override
  {
    RGWStorageStats bucket_stats;
    int ret = bucket_stats_cache.get_stats(user, bucket, bucket_stats,
                                           bucket_quota);
    if (ret < 0) {
      return ret;
    }

    if (bucket_stats.num_objects  + num_objs > num_shards * max_objs_per_shard) {
      ldout(store->ctx(), 0) << __func__ << ": resharding needed: stats.num_objects=" << bucket_stats.num_objects
             << " shard max_objects=" <<  max_objs_per_shard * num_shards << dendl;
      need_resharding = true;
      if (suggested_num_shards) {
        *suggested_num_shards = (bucket_stats.num_objects  + num_objs) * 2 / max_objs_per_shard;
      }
    } else {
      need_resharding = false;
    }

    return 0;
  }

};


RGWQuotaHandler *RGWQuotaHandler::generate_handler(RGWRados *store, bool quota_threads)
{
  return new RGWQuotaHandlerImpl(store, quota_threads);
}

void RGWQuotaHandler::free_handler(RGWQuotaHandler *handler)
{
  delete handler;
}


void rgw_apply_default_bucket_quota(RGWQuotaInfo& quota, const ConfigProxy& conf)
{
  if (conf->rgw_bucket_default_quota_max_objects >= 0) {
    quota.max_objects = conf->rgw_bucket_default_quota_max_objects;
    quota.enabled = true;
  }
  if (conf->rgw_bucket_default_quota_max_size >= 0) {
    quota.max_size = conf->rgw_bucket_default_quota_max_size;
    quota.enabled = true;
  }
}

void rgw_apply_default_user_quota(RGWQuotaInfo& quota, const ConfigProxy& conf)
{
  if (conf->rgw_user_default_quota_max_objects >= 0) {
    quota.max_objects = conf->rgw_user_default_quota_max_objects;
    quota.enabled = true;
  }
  if (conf->rgw_user_default_quota_max_size >= 0) {
    quota.max_size = conf->rgw_user_default_quota_max_size;
    quota.enabled = true;
  }
}