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
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
//! Kafka Client - A mid-level abstraction for a kafka cluster
//! allowing building higher level constructs.
//!
//! The entry point into this module is `KafkaClient` obtained by a
//! call to `KafkaClient::new()`.

use std;
use std::collections::hash_map;
use std::collections::hash_map::HashMap;
use std::io::Cursor;
use std::iter::Iterator;
use std::mem;
use std::thread;
use std::time::{Duration, Instant};

// pub re-export
pub use crate::compression::Compression;
pub use crate::utils::PartitionOffset;

#[cfg(feature = "security")]
pub use self::network::SecurityConfig;

use crate::codecs::{FromByte, ToByte};
use crate::error::{Error, KafkaCode, Result};
use crate::protocol::{self, ResponseParser};

use crate::client_internals::KafkaClientInternals;

pub mod metadata;
mod network;
mod state;

// ~ re-export (only) certain types from the protocol::fetch module as
// 'client::fetch'.
pub mod fetch {
    //! A representation of fetched messages from Kafka.

    pub use crate::protocol::fetch::{Data, Message, Partition, Response, Topic};
}

const DEFAULT_CONNECTION_RW_TIMEOUT_SECS: u64 = 120;

fn default_conn_rw_timeout() -> Option<Duration> {
    match DEFAULT_CONNECTION_RW_TIMEOUT_SECS {
        0 => None,
        n => Some(Duration::from_secs(n)),
    }
}

/// The default value for `KafkaClient::set_compression(..)`
pub const DEFAULT_COMPRESSION: Compression = Compression::NONE;

/// The default value for `KafkaClient::set_fetch_max_wait_time(..)`
pub const DEFAULT_FETCH_MAX_WAIT_TIME_MILLIS: u64 = 100;

/// The default value for `KafkaClient::set_fetch_min_bytes(..)`
pub const DEFAULT_FETCH_MIN_BYTES: i32 = 4096;

/// The default value for `KafkaClient::set_fetch_max_bytes(..)`
pub const DEFAULT_FETCH_MAX_BYTES_PER_PARTITION: i32 = 32 * 1024;

/// The default value for `KafkaClient::set_fetch_crc_validation(..)`
pub const DEFAULT_FETCH_CRC_VALIDATION: bool = true;

/// The default value for `KafkaClient::set_group_offset_storage(..)`
pub const DEFAULT_GROUP_OFFSET_STORAGE: Option<GroupOffsetStorage> = None;

/// The default value for `KafkaClient::set_retry_backoff_time(..)`
pub const DEFAULT_RETRY_BACKOFF_TIME_MILLIS: u64 = 100;

/// The default value for `KafkaClient::set_retry_max_attempts(..)`
// the default value: re-attempt a repeatable operation for
// approximetaly up to two minutes
pub const DEFAULT_RETRY_MAX_ATTEMPTS: u32 = 120_000 / DEFAULT_RETRY_BACKOFF_TIME_MILLIS as u32;

/// The default value for `KafkaClient::set_connection_idle_timeout(..)`
pub const DEFAULT_CONNECTION_IDLE_TIMEOUT_MILLIS: u64 = 540_000;

/// Client struct keeping track of brokers and topic metadata.
///
/// Implements methods described by the [Kafka Protocol](http://kafka.apache.org/protocol.html).
///
/// You will have to load metadata before making any other request.
#[derive(Debug)]
pub struct KafkaClient {
    // ~ this kafka client configuration
    config: ClientConfig,

    // ~ a pool of re-usable connections to kafka brokers
    conn_pool: network::Connections,

    // ~ the current state of this client
    state: state::ClientState,
}

#[derive(Debug)]
struct ClientConfig {
    client_id: String,
    hosts: Vec<String>,
    // ~ compression to use when sending messages
    compression: Compression,
    // ~ these are the defaults when fetching messages for details
    // refer to the kafka wire protocol
    fetch_max_wait_time: i32,
    fetch_min_bytes: i32,
    fetch_max_bytes_per_partition: i32,
    fetch_crc_validation: bool,
    // ~ the version of the API to use for the corresponding kafka
    // calls; note that this might have an effect on the storage type
    // kafka will then use (zookeeper or __consumer_offsets).  it is
    // important to use version for both of them which target the same
    // storage type.
    // offset_fetch_version: protocol::OffsetFetchVersion,
    // offset_commit_version: protocol::OffsetCommitVersion,
    offset_storage: Option<GroupOffsetStorage>,
    // ~ the duration to wait before retrying a failed
    // operation like refreshing group coordinators; this avoids
    // operation retries in a tight loop.
    retry_backoff_time: Duration,
    // ~ the number of repeated retry attempts; prevents endless
    // repetition of a retry attempt
    retry_max_attempts: u32,
}

// --------------------------------------------------------------------

/// Possible values when querying a topic's offset.
/// See `KafkaClient::fetch_offsets`.
#[derive(Debug, Copy, Clone)]
pub enum FetchOffset {
    /// Receive the earliest available offset.
    Earliest,
    /// Receive the latest offset.
    Latest,
    /// Used to ask for all messages before a certain time (ms); unix
    /// timestamp in milliseconds.
    /// See https://cwiki.apache.org/confluence/display/KAFKA/Writing+a+Driver+for+Kafka#WritingaDriverforKafka-Offsets
    ByTime(i64),
}

impl FetchOffset {
    fn to_kafka_value(self) -> i64 {
        match self {
            FetchOffset::Earliest => -2,
            FetchOffset::Latest => -1,
            FetchOffset::ByTime(n) => n,
        }
    }
}

// --------------------------------------------------------------------

/// Defines the available storage types to utilize when fetching or
/// committing group offsets.  See also `KafkaClient::set_group_offset_storage`.
#[derive(Debug, Copy, Clone, PartialEq, Eq)]
pub enum GroupOffsetStorage {
    /// Zookeeper based storage (available as of kafka 0.8.1)
    Zookeeper,
    /// Kafka based storage (available as of Kafka 0.8.2). This is the
    /// preferred method for groups to store their offsets at.
    Kafka,
}

impl GroupOffsetStorage {
    fn offset_fetch_version(self) -> protocol::OffsetFetchVersion {
        match self {
            GroupOffsetStorage::Zookeeper => protocol::OffsetFetchVersion::V0,
            GroupOffsetStorage::Kafka => protocol::OffsetFetchVersion::V1,
        }
    }
    fn offset_commit_version(self) -> protocol::OffsetCommitVersion {
        match self {
            GroupOffsetStorage::Zookeeper => protocol::OffsetCommitVersion::V0,
            // ~ if we knew we'll be communicating with a kafka 0.9+
            // broker we could set the commit-version to V2; however,
            // since we still want to support Kafka 0.8.2 versions,
            // we'll go with the less efficient but safe option V1.
            GroupOffsetStorage::Kafka => protocol::OffsetCommitVersion::V1,
        }
    }
}

/// Data point identifying a topic partition to fetch a group's offset
/// for.  See `KafkaClient::fetch_group_offsets`.
#[derive(Debug)]
pub struct FetchGroupOffset<'a> {
    /// The topic to fetch the group offset for
    pub topic: &'a str,
    /// The partition to fetch the group offset for
    pub partition: i32,
}

impl<'a> FetchGroupOffset<'a> {
    #[inline]
    pub fn new(topic: &'a str, partition: i32) -> Self {
        FetchGroupOffset { topic, partition }
    }
}

impl<'a> AsRef<FetchGroupOffset<'a>> for FetchGroupOffset<'a> {
    fn as_ref(&self) -> &Self {
        self
    }
}

// --------------------------------------------------------------------

/// Data point identifying a particular topic partition offset to be
/// committed.
/// See `KafkaClient::commit_offsets`.
#[derive(Debug)]
pub struct CommitOffset<'a> {
    /// The offset to be committed
    pub offset: i64,
    /// The topic to commit the offset for
    pub topic: &'a str,
    /// The partition to commit the offset for
    pub partition: i32,
}

impl<'a> CommitOffset<'a> {
    pub fn new(topic: &'a str, partition: i32, offset: i64) -> Self {
        CommitOffset {
            topic,
            partition,
            offset,
        }
    }
}

impl<'a> AsRef<CommitOffset<'a>> for CommitOffset<'a> {
    fn as_ref(&self) -> &Self {
        self
    }
}

// --------------------------------------------------------------------

/// Possible choices on acknowledgement requirements when
/// producing/sending messages to Kafka. See
/// `KafkaClient::produce_messages`.
#[derive(Debug, Copy, Clone)]
pub enum RequiredAcks {
    /// Indicates to the receiving Kafka broker not to acknowledge
    /// messages sent to it at all. Sending messages with this
    /// acknowledgement requirement translates into a fire-and-forget
    /// scenario which - of course - is very fast but not reliable.
    None = 0,
    /// Requires the receiving Kafka broker to wait until the sent
    /// messages are written to local disk.  Such messages can be
    /// regarded as acknowledged by one broker in the cluster.
    One = 1,
    /// Requires the sent messages to be acknowledged by all in-sync
    /// replicas of the targeted topic partitions.
    All = -1,
}

// --------------------------------------------------------------------

/// Message data to be sent/produced to a particular topic partition.
/// See `KafkaClient::produce_messages` and `Producer::send`.
#[derive(Debug)]
pub struct ProduceMessage<'a, 'b> {
    /// The "key" data of this message.
    pub key: Option<&'b [u8]>,

    /// The "value" data of this message.
    pub value: Option<&'b [u8]>,

    /// The topic to produce this message to.
    pub topic: &'a str,

    /// The partition (of the corresponding topic) to produce this
    /// message to.
    pub partition: i32,
}

impl<'a, 'b> AsRef<ProduceMessage<'a, 'b>> for ProduceMessage<'a, 'b> {
    fn as_ref(&self) -> &Self {
        self
    }
}

impl<'a, 'b> ProduceMessage<'a, 'b> {
    /// A convenient constructor method to create a new produce
    /// message with all attributes specified.
    pub fn new(
        topic: &'a str,
        partition: i32,
        key: Option<&'b [u8]>,
        value: Option<&'b [u8]>,
    ) -> Self {
        ProduceMessage {
            key,
            value,
            topic,
            partition,
        }
    }
}

// --------------------------------------------------------------------

/// Partition related request data for fetching messages.
/// See `KafkaClient::fetch_messages`.
#[derive(Debug)]
pub struct FetchPartition<'a> {
    /// The topic to fetch messages from.
    pub topic: &'a str,

    /// The offset as of which to fetch messages.
    pub offset: i64,

    /// The partition to fetch messasges from.
    pub partition: i32,

    /// Specifies the max. amount of data to fetch (for this
    /// partition.)  This implicitly defines the biggest message the
    /// client can accept.  If this value is too small, no messages
    /// can be delivered.  Setting this size should be in sync with
    /// the producers to the partition.
    ///
    /// Zero or negative values are treated as "unspecified".
    pub max_bytes: i32,
}

impl<'a> FetchPartition<'a> {
    /// Creates a new "fetch messages" request structure with an
    /// unspecified `max_bytes`.
    pub fn new(topic: &'a str, partition: i32, offset: i64) -> Self {
        FetchPartition {
            topic,
            partition,
            offset,
            max_bytes: -1,
        }
    }

    /// Sets the `max_bytes` value for the "fetch messages" request.
    pub fn with_max_bytes(mut self, max_bytes: i32) -> Self {
        self.max_bytes = max_bytes;
        self
    }
}

impl<'a> AsRef<FetchPartition<'a>> for FetchPartition<'a> {
    fn as_ref(&self) -> &Self {
        self
    }
}

/// A confirmation of messages sent back by the Kafka broker
/// to confirm delivery of producer messages.
#[derive(Debug)]
pub struct ProduceConfirm {
    /// The topic the messages were sent to.
    pub topic: String,

    /// The list of individual confirmations for each offset and partition.
    pub partition_confirms: Vec<ProducePartitionConfirm>,
}

/// A confirmation of messages sent back by the Kafka broker
/// to confirm delivery of producer messages for a particular topic.
#[derive(Debug)]
pub struct ProducePartitionConfirm {
    /// The offset assigned to the first message in the message set appended
    /// to this partition, or an error if one occurred.
    pub offset: std::result::Result<i64, KafkaCode>,

    /// The partition to which the message(s) were appended.
    pub partition: i32,
}

// --------------------------------------------------------------------

impl KafkaClient {
    /// Creates a new instance of KafkaClient. Before being able to
    /// successfully use the new client, you'll have to load metadata.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// let mut client = kafka::client::KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// client.load_metadata_all().unwrap();
    /// ```
    pub fn new(hosts: Vec<String>) -> KafkaClient {
        KafkaClient {
            config: ClientConfig {
                client_id: String::new(),
                hosts,
                compression: DEFAULT_COMPRESSION,
                fetch_max_wait_time: protocol::to_millis_i32(Duration::from_millis(
                    DEFAULT_FETCH_MAX_WAIT_TIME_MILLIS,
                ))
                .expect("invalid default-fetch-max-time-millis"),
                fetch_min_bytes: DEFAULT_FETCH_MIN_BYTES,
                fetch_max_bytes_per_partition: DEFAULT_FETCH_MAX_BYTES_PER_PARTITION,
                fetch_crc_validation: DEFAULT_FETCH_CRC_VALIDATION,
                offset_storage: DEFAULT_GROUP_OFFSET_STORAGE,
                retry_backoff_time: Duration::from_millis(DEFAULT_RETRY_BACKOFF_TIME_MILLIS),
                retry_max_attempts: DEFAULT_RETRY_MAX_ATTEMPTS,
            },
            conn_pool: network::Connections::new(
                default_conn_rw_timeout(),
                Duration::from_millis(DEFAULT_CONNECTION_IDLE_TIMEOUT_MILLIS),
            ),
            state: state::ClientState::new(),
        }
    }

    /// Creates a new secure instance of KafkaClient. Before being able to
    /// successfully use the new client, you'll have to load metadata.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// extern crate openssl;
    /// extern crate kafka;
    ///
    /// use openssl::ssl::{SslConnector, SslMethod, SslFiletype, SslVerifyMode};
    /// use kafka::client::{KafkaClient, SecurityConfig};
    ///
    /// fn main() {
    ///     let (key, cert) = ("client.key".to_string(), "client.crt".to_string());
    ///
    ///     // OpenSSL offers a variety of complex configurations. Here is an example:
    ///     let mut builder = SslConnector::builder(SslMethod::tls()).unwrap();
    ///     builder.set_cipher_list("DEFAULT").unwrap();
    ///     builder
    ///         .set_certificate_file(cert, SslFiletype::PEM)
    ///         .unwrap();
    ///     builder
    ///         .set_private_key_file(key, SslFiletype::PEM)
    ///         .unwrap();
    ///     builder.check_private_key().unwrap();
    ///     builder.set_default_verify_paths().unwrap();
    ///     builder.set_verify(SslVerifyMode::PEER);
    ///     let connector = builder.build();
    ///
    ///     let mut client = KafkaClient::new_secure(vec!("localhost:9092".to_owned()),
    ///                                              SecurityConfig::new(connector));
    ///     client.load_metadata_all().unwrap();
    /// }
    /// ```
    /// See also `SecurityConfig#with_hostname_verification` to disable hostname verification.
    ///
    /// See also `KafkaClient::load_metadatata_all` and
    /// `KafkaClient::load_metadata` methods, the creates
    /// [openssl](https://crates.io/crates/openssl)
    /// and [openssl_verify](https://crates.io/crates/openssl-verify),
    /// as well as
    /// [Kafka's documentation](https://kafka.apache.org/documentation.html#security_ssl).
    #[cfg(feature = "security")]
    pub fn new_secure(hosts: Vec<String>, security: SecurityConfig) -> KafkaClient {
        KafkaClient {
            config: ClientConfig {
                client_id: String::new(),
                hosts,
                compression: DEFAULT_COMPRESSION,
                fetch_max_wait_time: protocol::to_millis_i32(Duration::from_millis(
                    DEFAULT_FETCH_MAX_WAIT_TIME_MILLIS,
                ))
                .expect("invalid default-fetch-max-time-millis"),
                fetch_min_bytes: DEFAULT_FETCH_MIN_BYTES,
                fetch_max_bytes_per_partition: DEFAULT_FETCH_MAX_BYTES_PER_PARTITION,
                fetch_crc_validation: DEFAULT_FETCH_CRC_VALIDATION,
                offset_storage: DEFAULT_GROUP_OFFSET_STORAGE,
                retry_backoff_time: Duration::from_millis(DEFAULT_RETRY_BACKOFF_TIME_MILLIS),
                retry_max_attempts: DEFAULT_RETRY_MAX_ATTEMPTS,
            },
            conn_pool: network::Connections::new_with_security(
                default_conn_rw_timeout(),
                Duration::from_millis(DEFAULT_CONNECTION_IDLE_TIMEOUT_MILLIS),
                Some(security),
            ),
            state: state::ClientState::new(),
        }
    }

    /// Exposes the hosts used for discovery of the target kafka
    /// cluster.  This set of hosts corresponds to the values supplied
    /// to `KafkaClient::new`.
    #[inline]
    pub fn hosts(&self) -> &[String] {
        &self.config.hosts
    }

    /// Sets the client_id to be sent along every request to the
    /// remote Kafka brokers.  By default, this value is the empty
    /// string.
    ///
    /// Kafka brokers write out this client id to their
    /// request/response trace log - if configured appropriately.
    pub fn set_client_id(&mut self, client_id: String) {
        self.config.client_id = client_id;
    }

    /// Retrieves the current `KafkaClient::set_client_id` setting.
    pub fn client_id(&self) -> &str {
        &self.config.client_id
    }

    /// Sets the compression algorithm to use when sending out messages.
    ///
    /// # Example
    ///
    /// ```no_run
    /// use kafka::client::{Compression, KafkaClient};
    ///
    /// let mut client = KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// client.load_metadata_all().unwrap();
    /// client.set_compression(Compression::NONE);
    /// ```
    #[inline]
    pub fn set_compression(&mut self, compression: Compression) {
        self.config.compression = compression;
    }

    /// Retrieves the current `KafkaClient::set_compression` setting.
    #[inline]
    pub fn compression(&self) -> Compression {
        self.config.compression
    }

    /// Sets the maximum time in milliseconds to wait for insufficient
    /// data to become available when fetching messages.
    ///
    /// See also `KafkaClient::set_fetch_min_bytes(..)` and
    /// `KafkaClient::set_fetch_max_bytes_per_partition(..)`.
    #[inline]
    pub fn set_fetch_max_wait_time(&mut self, max_wait_time: Duration) -> Result<()> {
        self.config.fetch_max_wait_time = protocol::to_millis_i32(max_wait_time)?;
        Ok(())
    }

    /// Retrieves the current `KafkaClient::set_fetch_max_wait_time`
    /// setting.
    #[inline]
    pub fn fetch_max_wait_time(&self) -> Duration {
        Duration::from_millis(self.config.fetch_max_wait_time as u64)
    }

    /// Sets the minimum number of bytes of available data to wait for
    /// as long as specified by `KafkaClient::set_fetch_max_wait_time`
    /// when fetching messages.
    ///
    /// By setting higher values in combination with the timeout the
    /// consumer can tune for throughput and trade a little additional
    /// latency for reading only large chunks of data (e.g. setting
    /// MaxWaitTime to 100 ms and setting MinBytes to 64k would allow
    /// the server to wait up to 100ms to try to accumulate 64k of
    /// data before responding).
    ///
    /// # Example
    ///
    /// ```no_run
    /// use std::time::Duration;
    /// use kafka::client::{KafkaClient, FetchPartition};
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// client.set_fetch_max_wait_time(Duration::from_millis(100));
    /// client.set_fetch_min_bytes(64 * 1024);
    /// let r = client.fetch_messages(&[FetchPartition::new("my-topic", 0, 0)]);
    /// ```
    ///
    /// See also `KafkaClient::set_fetch_max_wait_time(..)` and
    /// `KafkaClient::set_fetch_max_bytes_per_partition(..)`.
    #[inline]
    pub fn set_fetch_min_bytes(&mut self, min_bytes: i32) {
        self.config.fetch_min_bytes = min_bytes;
    }

    /// Retrieves the current `KafkaClient::set_fetch_min_bytes`
    /// setting.
    #[inline]
    pub fn fetch_min_bytes(&self) -> i32 {
        self.config.fetch_min_bytes
    }

    /// Sets the default maximum number of bytes to obtain from _a
    /// single kafka partition_ when fetching messages.
    ///
    /// This basically determines the maximum message size this client
    /// will be able to fetch.  If a topic partition contains a
    /// message larger than this specified number of bytes, the server
    /// will not deliver it.
    ///
    /// Note that this setting is related to a single partition.  The
    /// overall potential data size in a fetch messages response will
    /// thus be determined by the number of partitions in the fetch
    /// messages request times this "max bytes per partitions."
    ///
    /// This client will use this setting by default for all queried
    /// partitions, however, `fetch_messages` does allow you to
    /// override this setting for a particular partition being
    /// queried.
    ///
    /// See also `KafkaClient::set_fetch_max_wait_time`,
    /// `KafkaClient::set_fetch_min_bytes`, and `KafkaClient::fetch_messages`.
    #[inline]
    pub fn set_fetch_max_bytes_per_partition(&mut self, max_bytes: i32) {
        self.config.fetch_max_bytes_per_partition = max_bytes;
    }

    /// Retrieves the current
    /// `KafkaClient::set_fetch_max_bytes_per_partition` setting.
    #[inline]
    pub fn fetch_max_bytes_per_partition(&self) -> i32 {
        self.config.fetch_max_bytes_per_partition
    }

    /// Specifies whether the to perform CRC validation on fetched
    /// messages.
    ///
    /// This ensures detection of on-the-wire or on-disk corruption to
    /// fetched messages.  This check adds some overhead, so it may be
    /// disabled in cases seeking extreme performance.
    #[inline]
    pub fn set_fetch_crc_validation(&mut self, validate_crc: bool) {
        self.config.fetch_crc_validation = validate_crc;
    }

    /// Retrieves the current `KafkaClient::set_fetch_crc_validation`
    /// setting.
    #[inline]
    pub fn fetch_crc_validation(&self) -> bool {
        self.config.fetch_crc_validation
    }

    /// Specifies the group offset storage to address when fetching or
    /// committing group offsets.
    ///
    /// In addition to Zookeeper, Kafka 0.8.2 brokers or later offer a
    /// more performant (and scalable) way to manage group offset
    /// directly by itself. Note that the remote storages are separate
    /// and independent on each other. Hence, you typically want
    /// consistently hard-code your choice in your program.
    ///
    /// Unless you have a 0.8.1 broker or want to participate in a
    /// group which is already based on Zookeeper, you generally want
    /// to choose `GroupOffsetStorage::Kafka` here.
    ///
    /// See also `KafkaClient::fetch_group_offsets` and
    /// `KafkaClient::commit_offsets`.
    #[inline]
    pub fn set_group_offset_storage(&mut self, storage: Option<GroupOffsetStorage>) {
        self.config.offset_storage = storage;
    }

    /// Retrieves the current `KafkaClient::set_group_offset_storage`
    /// settings.
    pub fn group_offset_storage(&self) -> Option<GroupOffsetStorage> {
        self.config.offset_storage
    }

    /// Specifies the time to wait before retrying a failed,
    /// repeatable operation against Kafka.  This avoids retrying such
    /// operations in a tight loop.
    #[inline]
    pub fn set_retry_backoff_time(&mut self, time: Duration) {
        self.config.retry_backoff_time = time;
    }

    /// Retrieves the current `KafkaClient::set_retry_backoff_time`
    /// setting.
    pub fn retry_backoff_time(&self) -> Duration {
        self.config.retry_backoff_time
    }

    /// Specifies the upper limit of retry attempts for failed,
    /// repeatable operations against kafka.  This avoids retrying
    /// them forever.
    #[inline]
    pub fn set_retry_max_attempts(&mut self, attempts: u32) {
        self.config.retry_max_attempts = attempts;
    }

    /// Retrieves the current `KafkaClient::set_retry_max_attempts`
    /// setting.
    #[inline]
    pub fn retry_max_attempts(&self) -> u32 {
        self.config.retry_max_attempts
    }

    /// Specifies the timeout after which idle connections will
    /// transparently be closed/re-established by `KafkaClient`.
    ///
    /// To be effective this value must be smaller than the [remote
    /// broker's `connections.max.idle.ms`
    /// setting](https://kafka.apache.org/documentation.html#brokerconfigs).
    #[inline]
    pub fn set_connection_idle_timeout(&mut self, timeout: Duration) {
        self.conn_pool.set_idle_timeout(timeout);
    }

    /// Retrieves the current
    /// `KafkaClient::set_connection_idle_timeout` setting.
    #[inline]
    pub fn connection_idle_timeout(&self) -> Duration {
        self.conn_pool.idle_timeout()
    }

    /// Provides a view onto the currently loaded metadata of known .
    ///
    /// # Examples
    /// ```no_run
    /// use kafka::client::KafkaClient;
    /// use kafka::client::metadata::Broker;
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// for topic in client.topics() {
    ///   for partition in topic.partitions() {
    ///     println!("{} #{} => {}", topic.name(), partition.id(),
    ///              partition.leader()
    ///                       .map(Broker::host)
    ///                       .unwrap_or("no-leader!"));
    ///   }
    /// }
    /// ```
    #[inline]
    pub fn topics(&self) -> metadata::Topics<'_> {
        metadata::Topics::new(self)
    }

    /// Resets and loads metadata for all topics from the underlying
    /// brokers.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// let mut client = kafka::client::KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// client.load_metadata_all().unwrap();
    /// for topic in client.topics().names() {
    ///   println!("topic: {}", topic);
    /// }
    /// ```
    ///
    /// Returns the metadata for all loaded topics underlying this
    /// client.
    #[inline]
    pub fn load_metadata_all(&mut self) -> Result<()> {
        self.reset_metadata();
        self.load_metadata::<&str>(&[])
    }

    /// Reloads metadata for a list of supplied topics.
    ///
    /// Note: if any of the specified topics does not exist yet on the
    /// underlying brokers and these have the [configuration for "auto
    /// create topics"
    /// enabled](https://kafka.apache.org/documentation.html#configuration),
    /// the remote kafka instance will create the yet missing topics
    /// on the fly as a result of explicitly loading their metadata.
    /// This is in contrast to other methods of this `KafkaClient`
    /// which will silently filter out requests to
    /// not-yet-loaded/not-yet-known topics and, thus, not cause
    /// topics to be automatically created.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// let mut client = kafka::client::KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// let _ = client.load_metadata(&["my-topic"]).unwrap();
    /// ```
    ///
    /// Returns the metadata for _all_ loaded topics underlying this
    /// client (this might be more topics than specified right to this
    /// method call.)
    #[inline]
    pub fn load_metadata<T: AsRef<str>>(&mut self, topics: &[T]) -> Result<()> {
        let resp = self.fetch_metadata(topics)?;
        self.state.update_metadata(resp)
    }

    /// Clears metadata stored in the client.  You must load metadata
    /// after this call if you want to use the client.
    #[inline]
    pub fn reset_metadata(&mut self) {
        self.state.clear_metadata();
    }

    /// Fetches metadata about the specified topics from all of the
    /// underlying brokers (`self.hosts`).
    fn fetch_metadata<T: AsRef<str>>(
        &mut self,
        topics: &[T],
    ) -> Result<protocol::MetadataResponse> {
        let correlation = self.state.next_correlation_id();
        let now = Instant::now();

        for host in &self.config.hosts {
            debug!("fetch_metadata: requesting metadata from {}", host);
            match self.conn_pool.get_conn(host, now) {
                Ok(conn) => {
                    let req =
                        protocol::MetadataRequest::new(correlation, &self.config.client_id, topics);
                    match __send_request(conn, req) {
                        Ok(_) => return __get_response::<protocol::MetadataResponse>(conn),
                        Err(e) => debug!(
                            "fetch_metadata: failed to request metadata from {}: {}",
                            host, e
                        ),
                    }
                }
                Err(e) => {
                    debug!("fetch_metadata: failed to connect to {}: {}", host, e);
                }
            }
        }
        Err(Error::NoHostReachable)
    }

    /// Fetch offsets for a list of topics
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::KafkaClient;
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// let topics: Vec<String> = client.topics().names().map(ToOwned::to_owned).collect();
    /// let offsets = client.fetch_offsets(&topics, kafka::client::FetchOffset::Latest).unwrap();
    /// ```
    ///
    /// Returns a mapping of topic name to `PartitionOffset`s for each
    /// currently available partition of the corresponding topic.
    pub fn fetch_offsets<T: AsRef<str>>(
        &mut self,
        topics: &[T],
        offset: FetchOffset,
    ) -> Result<HashMap<String, Vec<PartitionOffset>>> {
        let time = offset.to_kafka_value();
        let n_topics = topics.len();

        let state = &mut self.state;
        let correlation = state.next_correlation_id();

        // Map topic and partition to the corresponding broker
        let config = &self.config;
        let mut reqs: HashMap<&str, protocol::OffsetRequest<'_>> = HashMap::with_capacity(n_topics);
        for topic in topics {
            let topic = topic.as_ref();
            if let Some(ps) = state.partitions_for(topic) {
                for (id, host) in ps
                    .iter()
                    .filter_map(|(id, p)| p.broker(state).map(|b| (id, b.host())))
                {
                    let entry = reqs.entry(host).or_insert_with(|| {
                        protocol::OffsetRequest::new(correlation, &config.client_id)
                    });
                    entry.add(topic, id, time);
                }
            }
        }

        // Call each broker with the request formed earlier
        let now = Instant::now();
        let mut res: HashMap<String, Vec<PartitionOffset>> = HashMap::with_capacity(n_topics);
        for (host, req) in reqs {
            let resp =
                __send_receive::<_, protocol::OffsetResponse>(&mut self.conn_pool, host, now, req)?;
            for tp in resp.topic_partitions {
                let mut entry = res.entry(tp.topic);
                let mut new_resp_offsets = None;
                let mut err = None;
                // Use an explicit scope here to allow insertion into a vacant entry
                // below
                {
                    // Return a &mut to the response we will be collecting into to
                    // return from this function. If there are some responses we have
                    // already prepared, keep collecting into that; otherwise, make a
                    // new collection to return.
                    let resp_offsets = match entry {
                        hash_map::Entry::Occupied(ref mut e) => e.get_mut(),
                        hash_map::Entry::Vacant(_) => {
                            new_resp_offsets = Some(Vec::new());
                            new_resp_offsets.as_mut().unwrap()
                        }
                    };
                    for p in tp.partitions {
                        let partition_offset = match p.to_offset() {
                            Ok(po) => po,
                            Err(code) => {
                                err = Some((p.partition, code));
                                break;
                            }
                        };
                        resp_offsets.push(partition_offset);
                    }
                }
                if let Some((partition, code)) = err {
                    let topic = KafkaClient::get_key_from_entry(entry);
                    return Err(Error::TopicPartitionError {
                        topic_name: topic,
                        partition_id: partition,
                        error_code: code,
                    });
                }
                if let hash_map::Entry::Vacant(e) = entry {
                    // unwrap is ok because if it is Vacant, it would have
                    // been made into a Some above
                    e.insert(new_resp_offsets.unwrap());
                }
            }
        }

        Ok(res)
    }

    /// Takes ownership back from the given HashMap Entry.
    fn get_key_from_entry<'a, K: 'a, V: 'a>(entry: hash_map::Entry<'a, K, V>) -> K {
        match entry {
            hash_map::Entry::Occupied(e) => e.remove_entry().0,
            hash_map::Entry::Vacant(e) => e.into_key(),
        }
    }

    /// Fetch offset for a single topic.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::{KafkaClient, FetchOffset};
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// let offsets = client.fetch_topic_offsets("my-topic", FetchOffset::Latest).unwrap();
    /// ```
    ///
    /// Returns a vector of the offset data for each available partition.
    /// See also `KafkaClient::fetch_offsets`.
    pub fn fetch_topic_offsets<T: AsRef<str>>(
        &mut self,
        topic: T,
        offset: FetchOffset,
    ) -> Result<Vec<PartitionOffset>> {
        let topic = topic.as_ref();

        let mut m = self.fetch_offsets(&[topic], offset)?;
        let offs = m.remove(topic).unwrap_or_default();
        if offs.is_empty() {
            Err(Error::Kafka(KafkaCode::UnknownTopicOrPartition))
        } else {
            Ok(offs)
        }
    }

    /// Fetch messages from Kafka (multiple topic, partitions).
    ///
    /// It takes a vector specifying the topic partitions and their
    /// offsets as of which to fetch messages.  Additionally, the
    /// default "max fetch size per partition" can be explicitly
    /// overridden if it is "defined" - this is, if `max_bytes` is
    /// greater than zero.
    ///
    /// The result is exposed in a raw, complicated manner but allows
    /// for very efficient consumption possibilities. All of the data
    /// available through the returned fetch responses is bound to
    /// their lifetime as that data is merely a "view" into parts of
    /// the response structs.  If you need to keep individual messages
    /// for a longer time than the whole fetch responses, you'll need
    /// to make a copy of the message data.
    ///
    /// * This method transparently uncompresses messages (while Kafka
    /// might sent them in compressed format.)
    ///
    /// * This method ensures to skip messages with a lower offset
    /// than requested (while Kafka might for efficiency reasons sent
    /// messages with a lower offset.)
    ///
    /// Note: before using this method consider using
    /// `kafka::consumer::Consumer` instead which provides an easier
    /// to use API for the regular use-case of fetching messesage from
    /// Kafka.
    ///
    /// # Example
    ///
    /// This example demonstrates iterating all fetched messages from
    /// two topic partitions.  From one partition we allow Kafka to
    /// deliver to us the default number bytes as defined by
    /// `KafkaClient::set_fetch_max_bytes_per_partition`, from the
    /// other partition we allow Kafka to deliver up to 1MiB of
    /// messages.
    ///
    /// ```no_run
    /// use kafka::client::{KafkaClient, FetchPartition};
    ///
    /// let mut client = KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// client.load_metadata_all().unwrap();
    /// let reqs = &[FetchPartition::new("my-topic", 0, 0),
    ///              FetchPartition::new("my-topic-2", 0, 0).with_max_bytes(1024*1024)];
    /// let resps = client.fetch_messages(reqs).unwrap();
    /// for resp in resps {
    ///   for t in resp.topics() {
    ///     for p in t.partitions() {
    ///       match p.data() {
    ///         Err(ref e) => {
    ///           println!("partition error: {}:{}: {}", t.topic(), p.partition(), e)
    ///         }
    ///         Ok(ref data) => {
    ///           println!("topic: {} / partition: {} / latest available message offset: {}",
    ///                    t.topic(), p.partition(), data.highwatermark_offset());
    ///           for msg in data.messages() {
    ///             println!("topic: {} / partition: {} / message.offset: {} / message.len: {}",
    ///                      t.topic(), p.partition(), msg.offset, msg.value.len());
    ///           }
    ///         }
    ///       }
    ///     }
    ///   }
    /// }
    /// ```
    /// See also `kafka::consumer`.
    /// See also `KafkaClient::set_fetch_max_bytes_per_partition`.
    pub fn fetch_messages<'a, I, J>(&mut self, input: I) -> Result<Vec<fetch::Response>>
    where
        J: AsRef<FetchPartition<'a>>,
        I: IntoIterator<Item = J>,
    {
        let state = &mut self.state;
        let config = &self.config;

        let correlation = state.next_correlation_id();

        // Map topic and partition to the corresponding broker
        let mut reqs: HashMap<&str, protocol::FetchRequest<'_, '_>> = HashMap::new();
        for inp in input {
            let inp = inp.as_ref();
            if let Some(broker) = state.find_broker(inp.topic, inp.partition) {
                reqs.entry(broker)
                    .or_insert_with(|| {
                        protocol::FetchRequest::new(
                            correlation,
                            &config.client_id,
                            config.fetch_max_wait_time,
                            config.fetch_min_bytes,
                        )
                    })
                    .add(
                        inp.topic,
                        inp.partition,
                        inp.offset,
                        if inp.max_bytes > 0 {
                            inp.max_bytes
                        } else {
                            config.fetch_max_bytes_per_partition
                        },
                    );
            }
        }

        __fetch_messages(&mut self.conn_pool, config, reqs)
    }

    /// Fetch messages from a single kafka partition.
    ///
    /// See `KafkaClient::fetch_messages`.
    pub fn fetch_messages_for_partition<'a>(
        &mut self,
        req: &FetchPartition<'a>,
    ) -> Result<Vec<fetch::Response>> {
        self.fetch_messages(&[req])
    }

    /// Send a message to Kafka
    ///
    /// `required_acks` - indicates how many acknowledgements the
    /// servers should receive before responding to the request
    ///
    /// `ack_timeout` - provides a maximum time in milliseconds the
    /// server can await the receipt of the number of acknowledgements
    /// in `required_acks`
    ///
    /// `input` - the set of `ProduceMessage`s to send
    ///
    /// Note: Unlike the higher-level `Producer` API, this method will
    /// *not* automatically determine the partition to deliver the
    /// message to.  It will strictly try to send the message to the
    /// specified partition.
    ///
    /// Note: Trying to send messages to non-existing topics or
    /// non-existing partitions will result in an error.
    ///
    /// # Example
    ///
    /// ```no_run
    /// use std::time::Duration;
    /// use kafka::client::{KafkaClient, ProduceMessage, RequiredAcks};
    ///
    /// let mut client = KafkaClient::new(vec!("localhost:9092".to_owned()));
    /// client.load_metadata_all().unwrap();
    /// let req = vec![ProduceMessage::new("my-topic", 0, None, Some("a".as_bytes())),
    ///                ProduceMessage::new("my-topic-2", 0, None, Some("b".as_bytes()))];
    /// let resp = client.produce_messages(RequiredAcks::One, Duration::from_millis(100), req);
    /// println!("{:?}", resp);
    /// ```
    ///
    /// The return value will contain a vector of topic, partition,
    /// offset and error if any OR error:Error.

    // XXX rework signaling an error; note that we need to either return the
    // messages which kafka failed to accept or otherwise tell the client about them

    pub fn produce_messages<'a, 'b, I, J>(
        &mut self,
        acks: RequiredAcks,
        ack_timeout: Duration,
        messages: I,
    ) -> Result<Vec<ProduceConfirm>>
    where
        J: AsRef<ProduceMessage<'a, 'b>>,
        I: IntoIterator<Item = J>,
    {
        self.internal_produce_messages(acks as i16, protocol::to_millis_i32(ack_timeout)?, messages)
    }

    /// Commit offset for a topic partitions on behalf of a consumer group.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::{KafkaClient, CommitOffset};
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// client.commit_offsets("my-group",
    ///     &[CommitOffset::new("my-topic", 0, 100),
    ///       CommitOffset::new("my-topic", 1, 99)])
    ///    .unwrap();
    /// ```
    ///
    /// In this example, we commit the offset 100 for the topic
    /// partition "my-topic:0" and 99 for the topic partition
    /// "my-topic:1".  Once successfully committed, these can then be
    /// retrieved using `fetch_group_offsets` even from another
    /// process or at much later point in time to resume comusing the
    /// topic partitions as of these offsets.
    pub fn commit_offsets<'a, J, I>(&mut self, group: &str, offsets: I) -> Result<()>
    where
        J: AsRef<CommitOffset<'a>>,
        I: IntoIterator<Item = J>,
    {
        match self.config.offset_storage {
            Some(offset_storage) => {
                let mut req = protocol::OffsetCommitRequest::new(
                    group,
                    offset_storage.offset_commit_version(),
                    self.state.next_correlation_id(),
                    &self.config.client_id,
                );
                for o in offsets {
                    let o = o.as_ref();
                    if self.state.contains_topic_partition(o.topic, o.partition) {
                        req.add(o.topic, o.partition, o.offset, "");
                    } else {
                        return Err(Error::Kafka(KafkaCode::UnknownTopicOrPartition));
                    }
                }
                if req.topic_partitions.is_empty() {
                    debug!("commit_offsets: no offsets provided");
                    Ok(())
                } else {
                    __commit_offsets(req, &mut self.state, &mut self.conn_pool, &self.config)
                }
            }
            None => Err(Error::UnsetOffsetStorage),
        }
    }

    /// Commit offset of a particular topic partition on behalf of a
    /// consumer group.
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::KafkaClient;
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// client.commit_offset("my-group", "my-topic", 0, 100).unwrap();
    /// ```
    ///
    /// See also `KafkaClient::commit_offsets`.
    pub fn commit_offset(
        &mut self,
        group: &str,
        topic: &str,
        partition: i32,
        offset: i64,
    ) -> Result<()> {
        self.commit_offsets(group, &[CommitOffset::new(topic, partition, offset)])
    }

    /// Fetch offset for a specified list of topic partitions of a consumer group
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::{KafkaClient, FetchGroupOffset};
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    ///
    /// let offsets =
    ///      client.fetch_group_offsets("my-group",
    ///             &[FetchGroupOffset::new("my-topic", 0),
    ///               FetchGroupOffset::new("my-topic", 1)])
    ///             .unwrap();
    /// ```
    ///
    /// See also `KafkaClient::fetch_group_topic_offset`.
    pub fn fetch_group_offsets<'a, J, I>(
        &mut self,
        group: &str,
        partitions: I,
    ) -> Result<HashMap<String, Vec<PartitionOffset>>>
    where
        J: AsRef<FetchGroupOffset<'a>>,
        I: IntoIterator<Item = J>,
    {
        match self.config.offset_storage {
            Some(offset_storage) => {
                let mut req = protocol::OffsetFetchRequest::new(
                    group,
                    offset_storage.offset_fetch_version(),
                    self.state.next_correlation_id(),
                    &self.config.client_id,
                );
                for p in partitions {
                    let p = p.as_ref();
                    if self.state.contains_topic_partition(p.topic, p.partition) {
                        req.add(p.topic, p.partition);
                    } else {
                        return Err(Error::Kafka(KafkaCode::UnknownTopicOrPartition));
                    }
                }
                __fetch_group_offsets(req, &mut self.state, &mut self.conn_pool, &self.config)
            }
            None => Err(Error::UnsetOffsetStorage),
        }
    }

    /// Fetch offset for all partitions of a particular topic of a consumer group
    ///
    /// # Examples
    ///
    /// ```no_run
    /// use kafka::client::KafkaClient;
    ///
    /// let mut client = KafkaClient::new(vec!["localhost:9092".to_owned()]);
    /// client.load_metadata_all().unwrap();
    /// let offsets = client.fetch_group_topic_offset("my-group", "my-topic").unwrap();
    /// ```
    pub fn fetch_group_topic_offset(
        &mut self,
        group: &str,
        topic: &str,
    ) -> Result<Vec<PartitionOffset>> {
        match self.config.offset_storage {
            Some(offset_storage) => {
                let mut req = protocol::OffsetFetchRequest::new(
                    group,
                    offset_storage.offset_fetch_version(),
                    self.state.next_correlation_id(),
                    &self.config.client_id,
                );

                match self.state.partitions_for(topic) {
                    None => return Err(Error::Kafka(KafkaCode::UnknownTopicOrPartition)),
                    Some(tp) => {
                        for (id, _) in tp {
                            req.add(topic, id);
                        }
                    }
                }

                Ok(
                    __fetch_group_offsets(req, &mut self.state, &mut self.conn_pool, &self.config)?
                        .remove(topic)
                        .unwrap_or_default(),
                )
            }
            None => Err(Error::UnsetOffsetStorage),
        }

        // Ok(
        //     __fetch_group_offsets(req, &mut self.state, &mut self.conn_pool, &self.config)?
        //         .remove(topic)
        //         .unwrap_or_default(),
        // )
    }
}

impl KafkaClientInternals for KafkaClient {
    fn internal_produce_messages<'a, 'b, I, J>(
        &mut self,
        required_acks: i16,
        ack_timeout: i32,
        messages: I,
    ) -> Result<Vec<ProduceConfirm>>
    where
        J: AsRef<ProduceMessage<'a, 'b>>,
        I: IntoIterator<Item = J>,
    {
        let state = &mut self.state;
        let correlation = state.next_correlation_id();

        // ~ map topic and partition to the corresponding brokers
        let config = &self.config;
        let mut reqs: HashMap<&str, protocol::ProduceRequest<'_, '_>> = HashMap::new();
        for msg in messages {
            let msg = msg.as_ref();
            match state.find_broker(msg.topic, msg.partition) {
                None => return Err(Error::Kafka(KafkaCode::UnknownTopicOrPartition)),
                Some(broker) => reqs
                    .entry(broker)
                    .or_insert_with(|| {
                        protocol::ProduceRequest::new(
                            required_acks,
                            ack_timeout,
                            correlation,
                            &config.client_id,
                            config.compression,
                        )
                    })
                    .add(msg.topic, msg.partition, msg.key, msg.value),
            }
        }
        __produce_messages(&mut self.conn_pool, reqs, required_acks == 0)
    }
}

fn __get_group_coordinator<'a>(
    group: &str,
    state: &'a mut state::ClientState,
    conn_pool: &mut network::Connections,
    config: &ClientConfig,
    now: Instant,
) -> Result<&'a str> {
    if let Some(host) = state.group_coordinator(group) {
        // ~ decouple the lifetimes to make borrowck happy;
        // this is actually safe since we're immediately
        // returning this, so the follow up code is not
        // affected here
        return Ok(unsafe { mem::transmute(host) });
    }
    let correlation_id = state.next_correlation_id();
    let req = protocol::GroupCoordinatorRequest::new(group, correlation_id, &config.client_id);
    let mut attempt = 1;
    loop {
        // ~ idealy we'd make this work even if `load_metadata` has not
        // been called yet; if there are no connections available we can
        // try connecting to the user specified bootstrap server similar
        // to the way `load_metadata` works.
        let conn = conn_pool.get_conn_any(now).expect("available connection");
        debug!(
            "get_group_coordinator: asking for coordinator of '{}' on: {:?}",
            group, conn
        );
        let r = __send_receive_conn::<_, protocol::GroupCoordinatorResponse>(conn, &req)?;
        let retry_code = match r.into_result() {
            Ok(r) => {
                return Ok(state.set_group_coordinator(group, &r));
            }
            Err(Error::Kafka(e @ KafkaCode::GroupCoordinatorNotAvailable)) => e,
            Err(e) => {
                return Err(e);
            }
        };
        if attempt < config.retry_max_attempts {
            debug!(
                "get_group_coordinator: will retry request (c: {}) due to: {:?}",
                req.header.correlation_id, retry_code
            );
            attempt += 1;
            __retry_sleep(config);
        } else {
            return Err(Error::Kafka(retry_code));
        }
    }
}

fn __commit_offsets(
    req: protocol::OffsetCommitRequest<'_, '_>,
    state: &mut state::ClientState,
    conn_pool: &mut network::Connections,
    config: &ClientConfig,
) -> Result<()> {
    let mut attempt = 1;
    loop {
        let now = Instant::now();

        let tps = {
            let host = __get_group_coordinator(req.group, state, conn_pool, config, now)?;
            debug!(
                "__commit_offsets: sending offset commit request '{:?}' to: {}",
                req, host
            );
            __send_receive::<_, protocol::OffsetCommitResponse>(conn_pool, host, now, &req)?
                .topic_partitions
        };

        let mut retry_code = None;

        'rproc: for tp in tps {
            for p in tp.partitions {
                match p.to_error() {
                    None => {}
                    Some(e @ KafkaCode::GroupLoadInProgress) => {
                        retry_code = Some(e);
                        break 'rproc;
                    }
                    Some(e @ KafkaCode::NotCoordinatorForGroup) => {
                        debug!(
                            "commit_offsets: resetting group coordinator for '{}'",
                            req.group
                        );
                        state.remove_group_coordinator(req.group);
                        retry_code = Some(e);
                        break 'rproc;
                    }
                    Some(code) => {
                        // ~ immediately abort with the error
                        return Err(Error::Kafka(code));
                    }
                }
            }
        }
        match retry_code {
            Some(e) => {
                if attempt < config.retry_max_attempts {
                    debug!(
                        "commit_offsets: will retry request (c: {}) due to: {:?}",
                        req.header.correlation_id, e
                    );
                    attempt += 1;
                    __retry_sleep(config);
                }
            }
            None => {
                return Ok(());
            }
        }
    }
}

fn __fetch_group_offsets(
    req: protocol::OffsetFetchRequest<'_, '_, '_>,
    state: &mut state::ClientState,
    conn_pool: &mut network::Connections,
    config: &ClientConfig,
) -> Result<HashMap<String, Vec<PartitionOffset>>> {
    let mut attempt = 1;
    loop {
        let now = Instant::now();

        let r = {
            let host = __get_group_coordinator(req.group, state, conn_pool, config, now)?;
            debug!(
                "fetch_group_offsets: sending request {:?} to: {}",
                req, host
            );
            __send_receive::<_, protocol::OffsetFetchResponse>(conn_pool, host, now, &req)?
        };

        debug!("fetch_group_offsets: received response: {:#?}", r);

        let mut retry_code = None;
        let mut topic_map = HashMap::with_capacity(r.topic_partitions.len());

        'rproc: for tp in r.topic_partitions {
            let mut partition_offsets = Vec::with_capacity(tp.partitions.len());

            for p in tp.partitions {
                match p.get_offsets() {
                    Ok(o) => {
                        partition_offsets.push(o);
                    }
                    Err(Error::Kafka(e @ KafkaCode::GroupLoadInProgress)) => {
                        retry_code = Some(e);
                        break 'rproc;
                    }
                    Err(Error::Kafka(e @ KafkaCode::NotCoordinatorForGroup)) => {
                        debug!(
                            "fetch_group_offsets: resetting group coordinator for '{}'",
                            req.group
                        );
                        state.remove_group_coordinator(req.group);
                        retry_code = Some(e);
                        break 'rproc;
                    }
                    Err(e) => {
                        // ~ immediately abort with the error
                        return Err(e);
                    }
                }
            }

            topic_map.insert(tp.topic, partition_offsets);
        }

        // ~ have we processed the result successfully or shall we
        // retry once more?
        match retry_code {
            Some(e) => {
                if attempt < config.retry_max_attempts {
                    debug!(
                        "fetch_group_offsets: will retry request (c: {}) due to: {:?}",
                        req.header.correlation_id, e
                    );
                    attempt += 1;
                    __retry_sleep(config)
                } else {
                    return Err(Error::Kafka(e));
                }
            }
            None => {
                return Ok(topic_map);
            }
        }
    }
}

/// ~ carries out the given fetch requests and returns the response
fn __fetch_messages(
    conn_pool: &mut network::Connections,
    config: &ClientConfig,
    reqs: HashMap<&str, protocol::FetchRequest<'_, '_>>,
) -> Result<Vec<fetch::Response>> {
    let now = Instant::now();
    let mut res = Vec::with_capacity(reqs.len());
    for (host, req) in reqs {
        let p = protocol::fetch::ResponseParser {
            validate_crc: config.fetch_crc_validation,
            requests: Some(&req),
        };
        res.push(__z_send_receive(conn_pool, host, now, &req, &p)?);
    }
    Ok(res)
}

/// ~ carries out the given produce requests and returns the response
fn __produce_messages(
    conn_pool: &mut network::Connections,
    reqs: HashMap<&str, protocol::ProduceRequest<'_, '_>>,
    no_acks: bool,
) -> Result<Vec<ProduceConfirm>> {
    let now = Instant::now();
    if no_acks {
        for (host, req) in reqs {
            __send_noack::<_, protocol::ProduceResponse>(conn_pool, host, now, req)?;
        }
        Ok(vec![])
    } else {
        let mut res: Vec<ProduceConfirm> = vec![];
        for (host, req) in reqs {
            let resp = __send_receive::<_, protocol::ProduceResponse>(conn_pool, host, now, req)?;
            for tpo in resp.get_response() {
                res.push(tpo);
            }
        }
        Ok(res)
    }
}

fn __send_receive<T, V>(
    conn_pool: &mut network::Connections,
    host: &str,
    now: Instant,
    req: T,
) -> Result<V::R>
where
    T: ToByte,
    V: FromByte,
{
    __send_receive_conn::<T, V>(conn_pool.get_conn(host, now)?, req)
}

fn __send_receive_conn<T, V>(conn: &mut network::KafkaConnection, req: T) -> Result<V::R>
where
    T: ToByte,
    V: FromByte,
{
    __send_request(conn, req)?;
    __get_response::<V>(conn)
}

fn __send_noack<T, V>(
    conn_pool: &mut network::Connections,
    host: &str,
    now: Instant,
    req: T,
) -> Result<usize>
where
    T: ToByte,
    V: FromByte,
{
    let conn = conn_pool.get_conn(host, now)?;
    __send_request(conn, req)
}

fn __send_request<T: ToByte>(conn: &mut network::KafkaConnection, request: T) -> Result<usize> {
    // ~ buffer to receive data to be sent
    let mut buffer = Vec::with_capacity(4);
    // ~ reserve bytes for the actual request size (we'll fill in that later)
    buffer.extend_from_slice(&[0, 0, 0, 0]);
    // ~ encode the request data
    request.encode(&mut buffer)?;
    // ~ put the size of the request data into the reserved area
    let size = buffer.len() as i32 - 4;
    size.encode(&mut &mut buffer[..])?;

    trace!("__send_request: Sending bytes: {:?}", &buffer);

    // ~ send the prepared buffer
    conn.send(&buffer)
}

fn __get_response<T: FromByte>(conn: &mut network::KafkaConnection) -> Result<T::R> {
    let size = __get_response_size(conn)?;
    let resp = conn.read_exact_alloc(size as u64)?;

    trace!("__get_response: received bytes: {:?}", &resp);

    // {
    //     use std::fs::OpenOptions;
    //     use std::io::Write;
    //     let mut f = OpenOptions::new()
    //         .write(true)
    //         .truncate(true)
    //         .create(true)
    //         .open("/tmp/dump.dat")
    //         .unwrap();
    //     f.write_all(&resp[..]).unwrap();
    // }
    // ::super::protocol::
    // let thing = ::super::error::KafkaCode::from_protocol(self.error); // KafkaCode::decode_new::<T>(&mut Cursor::new(resp));

    T::decode_new(&mut Cursor::new(resp))
}

fn __z_send_receive<R, P>(
    conn_pool: &mut network::Connections,
    host: &str,
    now: Instant,
    req: R,
    parser: &P,
) -> Result<P::T>
where
    R: ToByte,
    P: ResponseParser,
{
    let conn = conn_pool.get_conn(host, now)?;
    __send_request(conn, req)?;
    __z_get_response(conn, parser)
}

fn __z_get_response<P>(conn: &mut network::KafkaConnection, parser: &P) -> Result<P::T>
where
    P: ResponseParser,
{
    let size = __get_response_size(conn)?;
    let resp = conn.read_exact_alloc(size as u64)?;

    // {
    //     use std::fs::OpenOptions;
    //     use std::io::Write;
    //     let mut f = OpenOptions::new()
    //         .write(true)
    //         .truncate(true)
    //         .create(true)
    //         .open("/tmp/dump.dat")
    //         .unwrap();
    //     f.write_all(&resp[..]).unwrap();
    // }

    parser.parse(resp)
}

fn __get_response_size(conn: &mut network::KafkaConnection) -> Result<i32> {
    let mut buf = [0u8; 4];
    conn.read_exact(&mut buf)?;
    i32::decode_new(&mut Cursor::new(&buf))
}

/// Suspends the calling thread for the configured "retry" time. This
/// method should be called _only_ as part of a retry attempt.
fn __retry_sleep(cfg: &ClientConfig) {
    thread::sleep(cfg.retry_backoff_time)
}