-
Notifications
You must be signed in to change notification settings - Fork 579
/
application.cc
2180 lines (1977 loc) · 82 KB
/
application.cc
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// Copyright 2020 Redpanda Data, Inc.
//
// Use of this software is governed by the Business Source License
// included in the file licenses/BSL.md
//
// As of the Change Date specified in that file, in accordance with
// the Business Source License, use of this software will be governed
// by the Apache License, Version 2.0
#include "redpanda/application.h"
#include "archival/fwd.h"
#include "archival/ntp_archiver_service.h"
#include "archival/upload_controller.h"
#include "archival/upload_housekeeping_service.h"
#include "cli_parser.h"
#include "cloud_storage/cache_service.h"
#include "cloud_storage/partition_recovery_manager.h"
#include "cloud_storage/remote.h"
#include "cloud_storage/topic_recovery_service.h"
#include "cloud_storage_clients/client_pool.h"
#include "cluster/bootstrap_service.h"
#include "cluster/cluster_discovery.h"
#include "cluster/cluster_utils.h"
#include "cluster/cluster_uuid.h"
#include "cluster/controller.h"
#include "cluster/ephemeral_credential_frontend.h"
#include "cluster/ephemeral_credential_service.h"
#include "cluster/fwd.h"
#include "cluster/id_allocator.h"
#include "cluster/id_allocator_frontend.h"
#include "cluster/members_manager.h"
#include "cluster/members_table.h"
#include "cluster/metadata_dissemination_handler.h"
#include "cluster/metadata_dissemination_service.h"
#include "cluster/node/local_monitor.h"
#include "cluster/node_isolation_watcher.h"
#include "cluster/node_status_rpc_handler.h"
#include "cluster/partition_balancer_rpc_handler.h"
#include "cluster/partition_manager.h"
#include "cluster/rm_partition_frontend.h"
#include "cluster/security_frontend.h"
#include "cluster/self_test_rpc_handler.h"
#include "cluster/service.h"
#include "cluster/topic_recovery_status_frontend.h"
#include "cluster/topic_recovery_status_rpc_handler.h"
#include "cluster/topics_frontend.h"
#include "cluster/tx_gateway.h"
#include "cluster/tx_gateway_frontend.h"
#include "cluster/types.h"
#include "compression/async_stream_zstd.h"
#include "compression/stream_zstd.h"
#include "config/configuration.h"
#include "config/endpoint_tls_config.h"
#include "config/node_config.h"
#include "config/seed_server.h"
#include "coproc/api.h"
#include "coproc/partition_manager.h"
#include "features/feature_table_snapshot.h"
#include "features/fwd.h"
#include "features/migrators.h"
#include "kafka/client/configuration.h"
#include "kafka/server/coordinator_ntp_mapper.h"
#include "kafka/server/fetch_session_cache.h"
#include "kafka/server/group_manager.h"
#include "kafka/server/group_router.h"
#include "kafka/server/queue_depth_monitor.h"
#include "kafka/server/quota_manager.h"
#include "kafka/server/rm_group_frontend.h"
#include "kafka/server/server.h"
#include "kafka/server/snc_quota_manager.h"
#include "kafka/server/usage_manager.h"
#include "model/fundamental.h"
#include "model/metadata.h"
#include "net/server.h"
#include "pandaproxy/rest/api.h"
#include "pandaproxy/schema_registry/api.h"
#include "raft/group_manager.h"
#include "raft/recovery_throttle.h"
#include "raft/service.h"
#include "redpanda/admin_server.h"
#include "resource_mgmt/io_priority.h"
#include "ssx/thread_worker.h"
#include "storage/backlog_controller.h"
#include "storage/chunk_cache.h"
#include "storage/compaction_controller.h"
#include "storage/directories.h"
#include "syschecks/syschecks.h"
#include "utils/file_io.h"
#include "utils/human.h"
#include "utils/uuid.h"
#include "version.h"
#include "vlog.h"
#include <seastar/core/abort_source.hh>
#include <seastar/core/memory.hh>
#include <seastar/core/metrics.hh>
#include <seastar/core/prometheus.hh>
#include <seastar/core/seastar.hh>
#include <seastar/core/sharded.hh>
#include <seastar/core/smp.hh>
#include <seastar/core/thread.hh>
#include <seastar/json/json_elements.hh>
#include <seastar/net/tls.hh>
#include <seastar/util/conversions.hh>
#include <seastar/util/defer.hh>
#include <seastar/util/log.hh>
#include <sys/resource.h>
#include <sys/utsname.h>
#include <chrono>
#include <exception>
#include <memory>
#include <vector>
// This file in the data directory tracks the metadata
// needed to detect crash loops.
static constexpr std::string_view crash_loop_tracker_file = "startup_log";
// Crash tracking resets every 1h.
static constexpr model::timestamp_clock::duration crash_reset_duration{1h};
static void set_local_kafka_client_config(
std::optional<kafka::client::configuration>& client_config,
const config::node_config& config) {
client_config.emplace();
const auto& kafka_api = config.kafka_api.value();
if (kafka_api.empty()) {
// No Kafka listeners configured, cannot configure
// a client.
return;
}
client_config->brokers.set_value(
std::vector<net::unresolved_address>{kafka_api[0].address});
const auto& kafka_api_tls = config::node().kafka_api_tls.value();
auto tls_it = std::find_if(
kafka_api_tls.begin(),
kafka_api_tls.end(),
[&kafka_api](const config::endpoint_tls_config& tls) {
return tls.name == kafka_api[0].name;
});
if (tls_it != kafka_api_tls.end()) {
client_config->broker_tls.set_value(tls_it->config);
}
}
static void
set_sr_kafka_client_defaults(kafka::client::configuration& client_config) {
if (!client_config.produce_batch_delay.is_overriden()) {
client_config.produce_batch_delay.set_value(0ms);
}
if (!client_config.produce_batch_record_count.is_overriden()) {
client_config.produce_batch_record_count.set_value(int32_t(0));
}
if (!client_config.produce_batch_size_bytes.is_overriden()) {
client_config.produce_batch_size_bytes.set_value(int32_t(0));
}
}
application::application(ss::sstring logger_name)
: _log(std::move(logger_name)){};
application::~application() = default;
void application::shutdown() {
// Stop accepting new requests.
if (_kafka_server.local_is_initialized()) {
_kafka_server.invoke_on_all(&net::server::shutdown_input).get();
}
if (_rpc.local_is_initialized()) {
_rpc.invoke_on_all(&rpc::rpc_server::shutdown_input).get();
}
// We schedule shutting down controller input and aborting its operation as
// one of the first shutdown steps. This way we terminate all long running
// operations before shutting down the RPC server, preventing it from
// waiting on background dispatch gate `close` call.
if (controller) {
controller->shutdown_input().get();
}
ss::do_for_each(
_migrators,
[](std::unique_ptr<features::feature_migrator>& fm) {
return fm->stop();
})
.get();
// Stop processing heartbeats before stopping the partition manager (and
// the underlying Raft consensus instances). Otherwise we'd process
// heartbeats for consensus objects that no longer exist.
if (raft_group_manager.local_is_initialized()) {
raft_group_manager.invoke_on_all(&raft::group_manager::stop_heartbeats)
.get();
}
if (topic_recovery_service.local_is_initialized()) {
topic_recovery_service
.invoke_on_all(
&cloud_storage::topic_recovery_service::shutdown_recovery)
.get();
}
// Stop any I/O to object store: this will cause any readers in flight
// to abort and enables partition shutdown to proceed reliably.
if (cloud_storage_clients.local_is_initialized()) {
cloud_storage_clients
.invoke_on_all(
&cloud_storage_clients::client_pool::shutdown_connections)
.get();
}
// Stop all partitions before destructing the subsystems (transaction
// coordinator, etc). This interrupts ongoing replication requests,
// allowing higher level state machines to shutdown cleanly.
if (partition_manager.local_is_initialized()) {
partition_manager
.invoke_on_all(&cluster::partition_manager::stop_partitions)
.get();
}
if (cp_partition_manager.local_is_initialized()) {
cp_partition_manager
.invoke_on_all(&coproc::partition_manager::stop_partitions)
.get();
}
// Wait for all requests to finish before destructing services that may be
// used by pending requests.
if (_kafka_server.local_is_initialized()) {
_kafka_server.invoke_on_all(&net::server::wait_for_shutdown).get();
_kafka_server.stop().get();
}
if (_kafka_conn_quotas.local_is_initialized()) {
_kafka_conn_quotas.stop().get();
}
if (_rpc.local_is_initialized()) {
_rpc.invoke_on_all(&rpc::rpc_server::wait_for_shutdown).get();
_rpc.stop().get();
}
// Shut down services in reverse order to which they were registered.
while (!_deferred.empty()) {
_deferred.pop_back();
}
}
static void log_system_resources(
ss::logger& log, const boost::program_options::variables_map& cfg) {
const auto shard_mem = ss::memory::stats();
auto total_mem = shard_mem.total_memory() * ss::smp::count;
/**
* IMPORTANT: copied out of seastar `resources.cc`, if logic in seastar will
* change we have to change our logic in here.
*/
const size_t default_reserve_memory = std::max<size_t>(
1536_MiB, 0.07 * total_mem);
auto reserve = cfg.contains("reserve-memory") ? ss::parse_memory_size(
cfg["reserve-memory"].as<std::string>())
: default_reserve_memory;
vlog(
log.info,
"System resources: {{ cpus: {}, available memory: {}, reserved memory: "
"{}}}",
ss::smp::count,
human::bytes(total_mem),
human::bytes(reserve));
struct rlimit nofile = {0, 0};
if (getrlimit(RLIMIT_NOFILE, &nofile) == 0) {
vlog(
log.info,
"File handle limit: {}/{}",
nofile.rlim_cur,
nofile.rlim_max);
} else {
vlog(log.warn, "Error {} querying file handle limit", errno);
}
}
namespace {
static constexpr std::string_view community_msg = R"banner(
Welcome to the Redpanda community!
Documentation: https://docs.redpanda.com - Product documentation site
GitHub Discussion: https://github.com/redpanda-data/redpanda/discussions - Longer, more involved discussions
GitHub Issues: https://github.com/redpanda-data/redpanda/issues - Report and track issues with the codebase
Support: https://support.redpanda.com - Contact the support team privately
Product Feedback: https://redpanda.com/feedback - Let us know how we can improve your experience
Slack: https://redpanda.com/slack - Chat about all things Redpanda. Join the conversation!
Twitter: https://twitter.com/redpandadata - All the latest Redpanda news!
)banner";
} // anonymous namespace
int application::run(int ac, char** av) {
std::setvbuf(stdout, nullptr, _IOLBF, 1024);
ss::app_template app(setup_app_config());
app.add_options()("version", po::bool_switch(), "print version and exit");
app.add_options()(
"redpanda-cfg",
po::value<std::string>(),
".yaml file config for redpanda");
// Validate command line args using options registered by the app and
// seastar. Keep the resulting variables in a temporary map so they don't
// live for the lifetime of the application.
{
po::variables_map vm;
if (!cli_parser{
ac,
av,
cli_parser::app_opts{app.get_options_description()},
cli_parser::ss_opts{app.get_conf_file_options_description()},
_log}
.validate_into(vm)) {
return 1;
}
if (vm["version"].as<bool>()) {
std::cout << redpanda_version() << std::endl;
return 0;
}
}
// use endl for explicit flushing
std::cout << community_msg << std::endl;
return app.run(ac, av, [this, &app] {
vlog(_log.info, "Redpanda {}", redpanda_version());
struct ::utsname buf;
::uname(&buf);
vlog(
_log.info,
"kernel={}, nodename={}, machine={}",
buf.release,
buf.nodename,
buf.machine);
auto& cfg = app.configuration();
log_system_resources(_log, cfg);
// NOTE: we validate required args here instead of above because run()
// catches some Seastar-specific args like --help that may result in
// valid omissions of required args.
validate_arguments(cfg);
return ss::async([this, &cfg] {
try {
::stop_signal app_signal;
auto deferred = ss::defer([this] {
shutdown();
vlog(_log.info, "Shutdown complete.");
});
// must initialize configuration before services
hydrate_config(cfg);
initialize();
check_environment();
check_for_crash_loop();
setup_metrics();
wire_up_and_start(app_signal);
post_start_tasks();
app_signal.wait().get();
vlog(_log.info, "Stopping...");
} catch (const ss::abort_requested_exception&) {
vlog(_log.info, "Redpanda startup aborted");
return 0;
} catch (...) {
vlog(
_log.error,
"Failure during startup: {}",
std::current_exception());
return 1;
}
return 0;
});
});
}
void application::initialize(
std::optional<YAML::Node> proxy_cfg,
std::optional<YAML::Node> proxy_client_cfg,
std::optional<YAML::Node> schema_reg_cfg,
std::optional<YAML::Node> schema_reg_client_cfg,
std::optional<scheduling_groups> groups) {
// Set up the abort_on_oom value based on the associated cluster config
// property, and watch for changes.
_abort_on_oom
= config::shard_local_cfg().memory_abort_on_alloc_failure.bind();
auto oom_config_watch = [this]() {
const bool value = (*_abort_on_oom)();
vlog(
_log.info,
"Setting abort_on_allocation_failure (abort on OOM): {}",
value);
ss::memory::set_abort_on_allocation_failure(value);
};
// execute the callback to apply the initial value
oom_config_watch();
_abort_on_oom->watch(oom_config_watch);
/*
* allocate per-core zstd decompression workspace and per-core
* async_stream_zstd workspaces. it can be several megabytes in size, so do
* it before memory becomes fragmented.
*/
ss::smp::invoke_on_all([] {
// TODO: remove this when stream_zstd is replaced with async_stream_zstd
// in v/kafka
compression::stream_zstd::init_workspace(
config::shard_local_cfg().zstd_decompress_workspace_bytes());
compression::initialize_async_stream_zstd(
config::shard_local_cfg().zstd_decompress_workspace_bytes());
}).get0();
if (config::shard_local_cfg().enable_pid_file()) {
syschecks::pidfile_create(config::node().pidfile_path());
}
smp_groups::config smp_groups_cfg{
.raft_group_max_non_local_requests
= config::shard_local_cfg().raft_smp_max_non_local_requests().value_or(
smp_groups::default_raft_non_local_requests(
config::shard_local_cfg().topic_partitions_per_shard()))};
smp_service_groups.create_groups(smp_groups_cfg).get();
_deferred.emplace_back(
[this] { smp_service_groups.destroy_groups().get(); });
if (groups) {
sched_groups = *groups;
return;
}
sched_groups.create_groups().get();
_scheduling_groups_probe.wire_up(sched_groups);
_deferred.emplace_back([this] {
_scheduling_groups_probe.clear();
sched_groups.destroy_groups().get();
});
if (proxy_cfg) {
_proxy_config.emplace(*proxy_cfg);
}
if (proxy_client_cfg) {
_proxy_client_config.emplace(*proxy_client_cfg);
}
if (schema_reg_cfg) {
_schema_reg_config.emplace(*schema_reg_cfg);
}
if (schema_reg_client_cfg) {
_schema_reg_client_config.emplace(*schema_reg_client_cfg);
}
}
void application::setup_metrics() {
setup_internal_metrics();
setup_public_metrics();
}
void application::setup_public_metrics() {
namespace sm = ss::metrics;
if (config::shard_local_cfg().disable_public_metrics()) {
return;
}
seastar::metrics::replicate_metric_families(
seastar::metrics::default_handle(),
{{"io_queue_total_read_ops", ssx::metrics::public_metrics_handle},
{"io_queue_total_write_ops", ssx::metrics::public_metrics_handle},
{"memory_allocated_memory", ssx::metrics::public_metrics_handle},
{"memory_free_memory", ssx::metrics::public_metrics_handle}})
.get();
_public_metrics.start().get();
const auto version_label = ssx::metrics::make_namespaced_label("version")(
redpanda_git_version());
const auto revision_label = ssx::metrics::make_namespaced_label("revision")(
redpanda_git_revision());
const auto build_labels = {version_label, revision_label};
_public_metrics
.invoke_on(
ss::this_shard_id(),
[build_labels](auto& public_metrics) {
public_metrics.groups.add_group(
"application",
{sm::make_gauge(
"uptime_seconds_total",
[] {
return std::chrono::duration<double>(ss::engine().uptime())
.count();
},
sm::description("Redpanda uptime in seconds"))
.aggregate({sm::shard_label}),
sm::make_gauge(
"build",
[] { return 1; },
sm::description("Redpanda build information"),
build_labels)
.aggregate({sm::shard_label})});
})
.get();
_public_metrics
.invoke_on_all([](auto& public_metrics) {
public_metrics.groups.add_group(
"cpu",
{sm::make_gauge(
"busy_seconds_total",
[] {
return std::chrono::duration<double>(
ss::engine().total_busy_time())
.count();
},
sm::description("Total CPU busy time in seconds"))});
})
.get();
_deferred.emplace_back([this] { _public_metrics.stop().get(); });
}
void application::setup_internal_metrics() {
namespace sm = ss::metrics;
if (config::shard_local_cfg().disable_metrics()) {
return;
}
// build info
auto version_label = sm::label("version");
auto revision_label = sm::label("revision");
std::vector<sm::label_instance> build_labels{
version_label(redpanda_git_version()),
revision_label(redpanda_git_revision()),
};
_metrics.add_group(
"application",
{
sm::make_gauge(
"uptime",
[] {
return std::chrono::duration_cast<std::chrono::milliseconds>(
ss::engine().uptime())
.count();
},
sm::description("Redpanda uptime in milliseconds")),
sm::make_gauge(
"build",
[] { return 1; },
sm::description("Redpanda build information"),
build_labels),
});
}
void application::validate_arguments(const po::variables_map& cfg) {
if (!cfg.count("redpanda-cfg")) {
throw std::invalid_argument("Missing redpanda-cfg flag");
}
}
ss::app_template::config application::setup_app_config() {
ss::app_template::config app_cfg;
app_cfg.name = "Redpanda";
using namespace std::literals::chrono_literals; // NOLINT
app_cfg.default_task_quota = 500us;
app_cfg.auto_handle_sigint_sigterm = false;
return app_cfg;
}
void application::hydrate_config(const po::variables_map& cfg) {
std::filesystem::path cfg_path(cfg["redpanda-cfg"].as<std::string>());
// Retain the original bytes loaded so that we can hexdump them later
// if YAML Parse fails.
// Related: https://github.com/redpanda-data/redpanda/issues/3798
auto yaml_raw_bytes = read_fully_tmpbuf(cfg_path).get0();
auto yaml_raw_str = ss::to_sstring(yaml_raw_bytes.clone());
YAML::Node config;
try {
config = YAML::Load(yaml_raw_str);
} catch (const YAML::ParserException& e) {
// For most parse errors we do not want to do a binary dump. For
// "unknown escape character" we dump it, to debug issue #3789 where
// apparently valid config files can cause this exception.
if (e.msg.find("unknown escape character") != std::string::npos) {
vlog(_log.error, "Dumping config on 'unknown escape character':");
iobuf iob;
iob.append(std::move(yaml_raw_bytes));
// A reasonable config file is usually only hundreds of bytes.
auto hexdump = iob.hexdump(16384);
vlog(_log.error, "{}", hexdump);
}
throw;
}
auto config_printer = [this](std::string_view service, const auto& cfg) {
std::vector<ss::sstring> items;
cfg.for_each([&items, &service](const auto& item) {
items.push_back(
ssx::sformat("{}.{}\t- {}", service, item, item.desc()));
});
std::sort(items.begin(), items.end());
for (const auto& item : items) {
vlog(_log.info, "{}", item);
}
};
ss::smp::invoke_on_all([&config, cfg_path] {
config::node().load(cfg_path, config);
}).get0();
auto node_config_errors = config::node().load(config);
for (const auto& i : node_config_errors) {
vlog(
_log.warn,
"Node property '{}' validation error: {}",
i.first,
i.second);
}
if (node_config_errors.size() > 0) {
throw std::invalid_argument("Validation errors in node config");
}
// This includes loading from local bootstrap file or legacy
// config file on first-start or upgrade cases.
_config_preload = cluster::config_manager::preload(config).get0();
vlog(_log.info, "Cluster configuration properties:");
vlog(_log.info, "(use `rpk cluster config edit` to change)");
config_printer("redpanda", config::shard_local_cfg());
vlog(_log.info, "Node configuration properties:");
vlog(_log.info, "(use `rpk redpanda config set <cfg> <value>` to change)");
config_printer("redpanda", config::node());
if (config["pandaproxy"]) {
_proxy_config.emplace(config["pandaproxy"]);
if (config["pandaproxy_client"]) {
_proxy_client_config.emplace(config["pandaproxy_client"]);
} else {
set_local_kafka_client_config(_proxy_client_config, config::node());
}
// override pandaparoxy_client.consumer_session_timeout_ms with
// pandaproxy.consumer_instance_timeout_ms
_proxy_client_config->consumer_session_timeout.set_value(
_proxy_config->consumer_instance_timeout.value());
config_printer("pandaproxy", *_proxy_config);
config_printer("pandaproxy_client", *_proxy_client_config);
}
if (config["schema_registry"]) {
_schema_reg_config.emplace(config["schema_registry"]);
if (config["schema_registry_client"]) {
_schema_reg_client_config.emplace(config["schema_registry_client"]);
} else {
set_local_kafka_client_config(
_schema_reg_client_config, config::node());
}
set_sr_kafka_client_defaults(*_schema_reg_client_config);
config_printer("schema_registry", *_schema_reg_config);
config_printer("schema_registry_client", *_schema_reg_client_config);
}
}
void application::check_environment() {
syschecks::systemd_message("checking environment (CPU, Mem)").get();
syschecks::cpu();
syschecks::memory(config::node().developer_mode());
storage::directories::initialize(
config::node().data_directory().as_sstring())
.get();
cloud_storage::cache::initialize(config::node().cloud_storage_cache_path())
.get();
if (config::shard_local_cfg().storage_strict_data_init()) {
// Look for the special file that indicates a user intends
// for the found data directory to be the one we use.
auto strict_data_dir_file
= config::node().strict_data_dir_file_path().string();
auto file_exists = ss::file_exists(strict_data_dir_file).get();
if (!file_exists) {
throw std::invalid_argument(ssx::sformat(
"Data directory not in expected state: {} not found, is the "
"expected filesystem mounted?",
strict_data_dir_file));
}
}
}
/// Here we check for too many consecutive unclean shutdowns/crashes
/// and abort the startup sequence if the limit exceeds
/// crash_loop_limit until the operator intervenes. Crash tracking
/// is reset if the node configuration changes or its been 1h since
/// the broker last failed to start. This metadata is tracked in the
/// tracker file. This is to prevent on disk state from piling up in
/// each unclean run and creating more state to recover for the next run.
void application::check_for_crash_loop() {
auto file_path = config::node().data_directory().path
/ crash_loop_tracker_file;
std::optional<crash_tracker_metadata> maybe_crash_md;
if (ss::file_exists(file_path.string()).get()) {
// Ok to read the entire file, it contains a serialized uint32_t.
auto buf = read_fully(file_path).get();
try {
maybe_crash_md = serde::from_iobuf<crash_tracker_metadata>(
std::move(buf));
} catch (const serde::serde_exception&) {
// A malformed log file, ignore and reset it later.
// We truncate it below.
vlog(_log.warn, "Ignorning malformed tracker file {}", file_path);
}
}
// Compute the checksum of the current node configuration.
auto current_config
= read_fully_to_string(config::node().get_cfg_file_path()).get0();
auto checksum = xxhash_64(current_config.c_str(), current_config.length());
if (maybe_crash_md) {
auto& crash_md = maybe_crash_md.value();
auto& limit = config::node().crash_loop_limit.value();
// Check if it has been atleast 1h since last unsuccessful restart.
// Tracking resets every 1h.
auto time_since_last_start
= model::duration_since_epoch(model::timestamp::now())
- model::duration_since_epoch(crash_md._last_start_ts);
auto crash_limit_ok = !limit || crash_md._crash_count <= limit.value();
auto node_config_changed = crash_md._config_checksum != checksum;
auto tracking_reset = time_since_last_start > crash_reset_duration;
auto ok_to_proceed = crash_limit_ok || node_config_changed
|| tracking_reset;
if (!ok_to_proceed) {
vlog(
_log.error,
"Crash loop detected. Too many consecutive crashes {}, exceeded "
"{} configured value {}. To recover Redpanda from this state, "
"manually remove file at path {}. Crash loop automatically "
"resets 1h after last crash or with node configuration changes.",
crash_md._crash_count,
config::node().crash_loop_limit.name(),
limit.value(),
file_path);
throw std::runtime_error("Crash loop detected, aborting startup.");
}
vlog(
_log.debug,
"Consecutive crashes detected: {} node config changed: {} "
"time based tracking reset: {}",
crash_md._crash_count,
node_config_changed,
tracking_reset);
if (node_config_changed || tracking_reset) {
crash_md._crash_count = 0;
}
}
// Truncate and bump the crash count. We consider a run to be unclean by
// default unless the scheduled cleanup (that runs very late in shutdown)
// resets the file. See schedule_crash_tracker_file_cleanup().
auto new_crash_count = maybe_crash_md
? maybe_crash_md.value()._crash_count + 1
: 1;
crash_tracker_metadata updated{
._crash_count = new_crash_count,
._config_checksum = checksum,
._last_start_ts = model::timestamp::now()};
write_fully(file_path, serde::to_iobuf(updated)).get();
ss::sync_directory(config::node().data_directory.value().as_sstring())
.get();
}
void application::schedule_crash_tracker_file_cleanup() {
// Schedule a deletion of the tracker file. On a clean shutdown,
// the tracker file should be deleted thus reseting the crash count on the
// next run. In case of an unclean shutdown, we already bumped
// the crash count and that should be taken into account in the
// next run.
// We emplace it in the front to make it the last task to run.
_deferred.emplace_front([&] {
auto file = config::node().data_directory().path
/ crash_loop_tracker_file;
ss::remove_file(file.string()).get();
ss::sync_directory(config::node().data_directory().as_sstring()).get();
vlog(_log.debug, "Deleted crash loop tracker file: {}", file);
});
}
static admin_server_cfg
admin_server_cfg_from_global_cfg(scheduling_groups& sgs) {
return admin_server_cfg{
.endpoints = config::node().admin(),
.endpoints_tls = config::node().admin_api_tls(),
.admin_api_docs_dir = config::node().admin_api_doc_dir(),
.sg = sgs.admin_sg()};
}
void application::configure_admin_server() {
if (config::node().admin().empty()) {
return;
}
syschecks::systemd_message("constructing http server").get();
construct_service(
_admin,
admin_server_cfg_from_global_cfg(sched_groups),
std::ref(partition_manager),
std::ref(cp_partition_manager),
controller.get(),
std::ref(shard_table),
std::ref(metadata_cache),
std::ref(_connection_cache),
std::ref(node_status_table),
std::ref(self_test_frontend),
std::ref(usage_manager),
_proxy.get(),
_schema_registry.get(),
std::ref(topic_recovery_service),
std::ref(topic_recovery_status_frontend))
.get();
}
static storage::kvstore_config kvstore_config_from_global_config() {
/*
* The key-value store is rooted at the configured data directory, and
* the internal kvstore topic-namespace results in a storage layout of:
*
* /var/lib/redpanda/data/
* - redpanda/kvstore/
* - 0
* - 1
* - ... #cores
*/
return storage::kvstore_config(
config::shard_local_cfg().kvstore_max_segment_size(),
config::shard_local_cfg().kvstore_flush_interval.bind(),
config::node().data_directory().as_sstring(),
storage::debug_sanitize_files::no);
}
static storage::log_config
manager_config_from_global_config(scheduling_groups& sgs) {
return storage::log_config(
config::node().data_directory().as_sstring(),
config::shard_local_cfg().log_segment_size.bind(),
config::shard_local_cfg().compacted_log_segment_size.bind(),
config::shard_local_cfg().max_compacted_log_segment_size.bind(),
storage::jitter_percents(
config::shard_local_cfg().log_segment_size_jitter_percent()),
storage::debug_sanitize_files::no,
priority_manager::local().compaction_priority(),
config::shard_local_cfg().retention_bytes.bind(),
config::shard_local_cfg().log_compaction_interval_ms.bind(),
config::shard_local_cfg().delete_retention_ms.bind(),
storage::with_cache(!config::shard_local_cfg().disable_batch_cache()),
storage::batch_cache::reclaim_options{
.growth_window = config::shard_local_cfg().reclaim_growth_window(),
.stable_window = config::shard_local_cfg().reclaim_stable_window(),
.min_size = config::shard_local_cfg().reclaim_min_size(),
.max_size = config::shard_local_cfg().reclaim_max_size(),
.min_free_memory
= config::shard_local_cfg().reclaim_batch_cache_min_free(),
},
config::shard_local_cfg().readers_cache_eviction_timeout_ms(),
sgs.compaction_sg());
}
static storage::backlog_controller_config compaction_controller_config(
ss::scheduling_group sg, const ss::io_priority_class& iopc) {
auto space_info = std::filesystem::space(
config::node().data_directory().path);
/**
* By default we set desired compaction backlog size to 10% of disk
* capacity.
*/
static const int64_t backlog_capacity_percents = 10;
int64_t backlog_size
= config::shard_local_cfg().compaction_ctrl_backlog_size().value_or(
(space_info.capacity / 100) * backlog_capacity_percents
/ ss::smp::count);
/**
* We normalize internals using disk capacity to make controller settings
* independent from disk space. After normalization all values equal to disk
* capacity will be represented in the controller with value equal 1000.
*
* Set point = 10% of disk capacity will always be equal to 100.
*
* This way we can calculate proportional coefficient.
*
* We assume that when error is greater than 80% of setpoint we should be
* running compaction with maximum allowed shares.
* This way we can calculate proportional coefficient as
*
* k_p = 1000 / 80 = 12.5
*
*/
auto normalization = space_info.capacity / (1000 * ss::smp::count);
return storage::backlog_controller_config(
config::shard_local_cfg().compaction_ctrl_p_coeff(),
config::shard_local_cfg().compaction_ctrl_i_coeff(),
config::shard_local_cfg().compaction_ctrl_d_coeff(),
normalization,
backlog_size,
200,
config::shard_local_cfg().compaction_ctrl_update_interval_ms(),
sg,
iopc,
config::shard_local_cfg().compaction_ctrl_min_shares(),
config::shard_local_cfg().compaction_ctrl_max_shares());
}
static storage::backlog_controller_config
make_upload_controller_config(ss::scheduling_group sg) {
// This settings are similar to compaction_controller_config.
// The desired setpoint for archival is set to 0 since the goal is to upload
// all data that we have.
// If the size of the backlog (the data which should be uploaded to S3) is
// larger than this value we need to bump the scheduling priority.
// Otherwise, we're good with the minimal.
// Since the setpoint is 0 we can't really use integral component of the
// controller. This is because upload backlog size never gets negative so
// once integral part will rump up high enough it won't be able to go down
// even if everything is uploaded.
auto available
= ss::fs_avail(config::node().data_directory().path.string()).get();
int64_t setpoint = 0;
int64_t normalization = static_cast<int64_t>(available)
/ (1000 * ss::smp::count);
return {
config::shard_local_cfg().cloud_storage_upload_ctrl_p_coeff(),
0,
config::shard_local_cfg().cloud_storage_upload_ctrl_d_coeff(),
normalization,
setpoint,
static_cast<int>(
priority_manager::local().archival_priority().get_shares()),
config::shard_local_cfg().cloud_storage_upload_ctrl_update_interval_ms(),
sg,
priority_manager::local().archival_priority(),
config::shard_local_cfg().cloud_storage_upload_ctrl_min_shares(),
config::shard_local_cfg().cloud_storage_upload_ctrl_max_shares()};
}
// add additional services in here
void application::wire_up_runtime_services(model::node_id node_id) {
wire_up_redpanda_services(node_id);
if (_proxy_config) {
construct_single_service(
_proxy,
smp_service_groups.proxy_smp_sg(),
// TODO: Improve memory budget for services
// https://github.com/redpanda-data/redpanda/issues/1392
memory_groups::kafka_total_memory(),
*_proxy_client_config,
*_proxy_config,
controller.get());
}
if (_schema_reg_config) {
construct_single_service(
_schema_registry,
node_id,
smp_service_groups.proxy_smp_sg(),
// TODO: Improve memory budget for services
// https://github.com/redpanda-data/redpanda/issues/1392
memory_groups::kafka_total_memory(),
*_schema_reg_client_config,
*_schema_reg_config,
std::reference_wrapper(controller));
}
construct_single_service(_monitor_unsafe_log_flag, std::ref(feature_table));
configure_admin_server();
}
void application::wire_up_redpanda_services(model::node_id node_id) {
ss::smp::invoke_on_all([] {
resources::available_memory::local().register_metrics();
return storage::internal::chunks().start();
}).get();
construct_single_service(thread_worker);
// cluster