AK
можно настроить, чтобы в topology.yaml прописывалось dns-имя, а не ip?
Size: a a a
2020 August 06
YD
Можно
YD
основной посыл advertise_uri — "обращайтесь ко мне по такому-то урлу"
DI
основной посыл advertise_uri — "обращайтесь ко мне по такому-то урлу"
одна нода
первый старт
первый старт
Starting imdb_imdb_1 ... done
Attaching to imdb_imdb_1
imdb_1 | Auto-detected IP to be "172.31.0.2"
imdb_1 | Using advertise_uri "172.31.0.2:3301"
imdb_1 | Membership encryption enabled
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3302
imdb_1 | Membership BROADCAST sent to 172.31.255.255:3302
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3301
imdb_1 | Membership BROADCAST sent to 172.31.255.255:3301
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3300
imdb_1 | Membership BROADCAST sent to 172.31.255.255:3300
imdb_1 | started
imdb_1 | Listening HTTP on 0.0.0.0:8081
imdb_1 | started
imdb_1 | Remote control bound to 0.0.0.0:3301
imdb_1 | Remote control ready to accept connections
imdb_1 | Instance state changed: -> Unconfigured
imdb_1 | tcp_server: remove dead UNIX socket: /var/run/tarantool/imdb.imdb-router.control
imdb_1 | started
imdb_1 | entering the event loop
imdb_1 | Updating config clusterwide...
imdb_1 | (2PC) Preparation stage...
imdb_1 | Prepared for config update at 172.31.0.2:3301
imdb_1 | (2PC) Commit stage...
imdb_1 | Instance state changed: Unconfigured -> BootstrappingBox
imdb_1 | Calling box.cfg()...
imdb_1 | 2020-08-06 08:59:22.880 [1] main/112/remote_control/172.31.0.2:46206 C> Tarantool 2.2.3-1-g98ecc909a
imdb_1 | 2020-08-06 08:59:22.880 [1] main/112/remote_control/172.31.0.2:46206 C> log level 5
imdb_1 | 2020-08-06 08:59:22.883 [1] main/112/remote_control/172.31.0.2:46206 I> mapping 268435456 bytes for memtx tuple arena...
imdb_1 | 2020-08-06 08:59:22.883 [1] main/112/remote_control/172.31.0.2:46206 I> mapping 134217728 bytes for vinyl tuple arena...
imdb_1 | 2020-08-06 08:59:22.885 [1] main/112/remote_control/172.31.0.2:46206 I> instance uuid 737f3166-f9df-4032-ac1f-9e8652b15a80
imdb_1 | 2020-08-06 08:59:22.885 [1] main/112/remote_control/172.31.0.2:46206 I> initializing an empty data directory
imdb_1 | 2020-08-06 08:59:22.894 [1] main/112/remote_control/172.31.0.2:46206 I> assigned id 1 to replica 737f3166-f9df-4032-ac1f-9e8652b15a80
imdb_1 | 2020-08-06 08:59:22.894 [1] main/112/remote_control/172.31.0.2:46206 I> cluster uuid 7edd1f6b-dde9-4205-88f1-384d1272a3fb
imdb_1 | 2020-08-06 08:59:22.899 [1] snapshot/101/main I> saving snapshot `/var/lib/tarantool/imdb.imdb-router/00000000000000000000.snap.inprogress'
imdb_1 | 2020-08-06 08:59:22.904 [1] snapshot/101/main I> done
imdb_1 | 2020-08-06 08:59:22.907 [1] main/112/remote_control/172.31.0.2:46206 I> ready to accept requests
imdb_1 | 2020-08-06 08:59:22.907 [1] main/112/remote_control/172.31.0.2:46206 I> set 'custom_proc_title' configuration option to "imdb@imdb-router"
imdb_1 | 2020-08-06 08:59:22.907 [1] main/116/checkpoint_daemon I> scheduled next checkpoint for Thu Aug 6 10:15:01 2020
imdb_1 | 2020-08-06 08:59:22.908 [1] main/112/remote_control/172.31.0.2:46206 I> set 'instance_uuid' configuration option to "737f3166-f9df-4032-ac1f-9e8652b15a80"
imdb_1 | 2020-08-06 08:59:22.908 [1] main/112/remote_control/172.31.0.2:46206 I> set 'replication_connect_quorum' configuration option to 0
imdb_1 | 2020-08-06 08:59:22.908 [1] main/112/remote_control/172.31.0.2:46206 I> set 'replicaset_uuid' configuration option to "7edd1f6b-dde9-4205-88f1-384d1272a3fb"
imdb_1 | 2020-08-06 08:59:22.908 [1] main/112/remote_control/172.31.0.2:46206 I> Making sure user "admin" exists...
imdb_1 | 2020-08-06 08:59:22.908 [1] main/112/remote_control/172.31.0.2:46206 I> Granting replication permissions to "admin"...
imdb_1 | 2020-08-06 08:59:22.913 [1] wal/101/main xlog.c:1020 W> fallocate is not supported, proceeding without it
imdb_1 | 2020-08-06 08:59:22.913 [1] main/112/remote_control/172.31.0.2:46206 I> Setting password for user "admin" ...
imdb_1 | 2020-08-06 08:59:22.914 [1] iproto/101/main I> binary: bound to 0.0.0.0:3301
imdb_1 | 2020-08-06 08:59:22.914 [1] main/107/remote_control/0.0.0.0:3301 utils.c:1018 E> LuajitError: builtin/socket.lua:83: attempt to use closed socket
DI
основной посыл advertise_uri — "обращайтесь ко мне по такому-то урлу"
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> set 'listen' configuration option to "3301"
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> Instance state changed: BootstrappingBox -> ConnectingFullmesh
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> connecting to 1 replicas
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 C> failed to connect to 1 out of 1 replicas
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 C> leaving orphan mode
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> set 'replication' configuration option to ["admin@172.31.0.2:3301"]
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> Instance state changed: ConnectingFullmesh -> BoxConfigured
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> Instance state changed: BoxConfigured -> ConfiguringRoles
imdb_1 | 2020-08-06 08:59:22.914 [1] main/112/remote_control/172.31.0.2:46206 I> Failover disabled
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> Reconfiguring vshard.storage...
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> Starting configuration of replica 737f3166-f9df-4032-ac1f-9e8652b15a80
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> I am master
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> Taking on replicaset master role...
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> Box has been configured
imdb_1 | 2020-08-06 08:59:22.915 [1] main/112/remote_control/172.31.0.2:46206 I> Initializing schema {0.1.15.0}
imdb_1 | 2020-08-06 08:59:22.915 [1] main/128/applier/admin@172.31.0.2:3301 I> remote master 737f3166-f9df-4032-ac1f-9e8652b15a80 at 172.31.0.2:3301 running Tarantool 2.2.3
imdb_1 | 2020-08-06 08:59:22.916 [1] main/128/applier/admin@172.31.0.2:3301 C> leaving orphan mode
imdb_1 | 2020-08-06 08:59:22.924 [1] main/112/remote_control/172.31.0.2:46206 I> Upgrade vshard schema to {0.1.16.0}
imdb_1 | 2020-08-06 08:59:22.924 [1] main/112/remote_control/172.31.0.2:46206 I> Insert 'vshard_version' into _schema
imdb_1 | 2020-08-06 08:59:22.924 [1] main/112/remote_control/172.31.0.2:46206 I> Create function vshard.storage._call()
imdb_1 | 2020-08-06 08:59:22.925 [1] main/112/remote_control/172.31.0.2:46206 I> Successful vshard schema upgrade to {0.1.16.0}
imdb_1 | 2020-08-06 08:59:22.925 [1] main/129/lua I> gc_bucket_f has been started
imdb_1 | 2020-08-06 08:59:22.925 [1] main/130/lua I> recovery_f has been started
imdb_1 | 2020-08-06 08:59:22.925 [1] main/112/remote_control/172.31.0.2:46206 I> Took on replicaset master role
imdb_1 | 2020-08-06 08:59:22.925 [1] main/131/lua I> rebalancer_f has been started
imdb_1 | 2020-08-06 08:59:22.931 [1] main/112/remote_control/172.31.0.2:46206 I> Reconfiguring vshard-router/default ...
imdb_1 | 2020-08-06 08:59:22.931 [1] main/112/remote_control/172.31.0.2:46206 I> Starting router configuration
imdb_1 | 2020-08-06 08:59:22.931 [1] main/112/remote_control/172.31.0.2:46206 I> Calling box.cfg()...
imdb_1 | 2020-08-06 08:59:22.931 [1] main/112/remote_control/172.31.0.2:46206 I> {"read_only":false}
imdb_1 | 2020-08-06 08:59:22.931 [1] main/112/remote_control/172.31.0.2:46206 I> Box has been configured
imdb_1 | 2020-08-06 08:59:22.933 [1] main/132/172.31.0.2:3301 (net.box) I> connected to 172.31.0.2:3301
imdb_1 | 2020-08-06 08:59:22.933 [1] main/133/172.31.0.2:3301 (net.box) I> connected to 172.31.0.2:3301
Назначился ip 172.31.0.2, он же прописался в topology.yaml
DI
DI
второй старт
DI
Creating imdb_imdb_1 ... done
Attaching to imdb_imdb_1
imdb_1 | Auto-detected IP to be "192.168.0.2"
imdb_1 | Using advertise_uri "192.168.0.2:3301"
imdb_1 | Membership encryption enabled
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3302
imdb_1 | Membership BROADCAST sent to 192.168.15.255:3302
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3301
imdb_1 | Membership BROADCAST sent to 192.168.15.255:3301
imdb_1 | Membership BROADCAST sent to 127.0.0.1:3300
imdb_1 | Membership BROADCAST sent to 192.168.15.255:3300
imdb_1 | started
imdb_1 | Listening HTTP on 0.0.0.0:8081
imdb_1 | started
imdb_1 | Remote control bound to 0.0.0.0:3301
imdb_1 | Instance state changed: -> ConfigFound
imdb_1 | Instance state changed: ConfigFound -> ConfigLoaded
imdb_1 | started
imdb_1 | Instance state changed: ConfigLoaded -> RecoveringSnapshot
imdb_1 | Calling box.cfg()...
imdb_1 | 2020-08-06 08:59:34.172 [1] main/108/lua C> Tarantool 2.2.3-1-g98ecc909a
imdb_1 | 2020-08-06 08:59:34.172 [1] main/108/lua C> log level 5
imdb_1 | 2020-08-06 08:59:34.177 [1] main/108/lua I> mapping 268435456 bytes for memtx tuple arena...
imdb_1 | 2020-08-06 08:59:34.177 [1] main/108/lua I> mapping 134217728 bytes for vinyl tuple arena...
imdb_1 | 2020-08-06 08:59:34.179 [1] main C> entering the event loop
imdb_1 | 2020-08-06 08:59:34.180 [1] main/108/lua I> instance uuid 737f3166-f9df-4032-ac1f-9e8652b15a80
imdb_1 | 2020-08-06 08:59:34.186 [1] main/108/lua I> instance vclock {1: 30043}
imdb_1 | 2020-08-06 08:59:34.188 [1] main/108/lua I> recovery start
imdb_1 | 2020-08-06 08:59:34.188 [1] main/108/lua I> recovering from `/var/lib/tarantool/imdb.imdb-router/00000000000000000000.snap'
imdb_1 | 2020-08-06 08:59:34.190 [1] main/108/lua I> cluster uuid 7edd1f6b-dde9-4205-88f1-384d1272a3fb
imdb_1 | 2020-08-06 08:59:34.195 [1] main/108/lua I> assigned id 1 to replica 737f3166-f9df-4032-ac1f-9e8652b15a80
imdb_1 | 2020-08-06 08:59:34.198 [1] main/108/lua I> recover from `/var/lib/tarantool/imdb.imdb-router/00000000000000000000.xlog'
imdb_1 | 2020-08-06 08:59:34.219 [1] main/108/lua I> done `/var/lib/tarantool/imdb.imdb-router/00000000000000000000.xlog'
imdb_1 | 2020-08-06 08:59:34.222 [1] main/108/lua I> recover from `/var/lib/tarantool/imdb.imdb-router/00000000000000030043.xlog'
imdb_1 | 2020-08-06 08:59:34.222 [1] main/108/lua I> done `/var/lib/tarantool/imdb.imdb-router/00000000000000030043.xlog'
imdb_1 | 2020-08-06 08:59:34.222 [1] main/108/lua I> Building secondary indexes in space '_bucket'...
imdb_1 | 2020-08-06 08:59:34.222 [1] main/108/lua I> Adding 30000 keys to TREE index 'status' ...
imdb_1 | 2020-08-06 08:59:34.225 [1] main/108/lua I> Space '_bucket': done
imdb_1 | 2020-08-06 08:59:34.233 [1] main/108/lua I> ready to accept requests
imdb_1 | 2020-08-06 08:59:34.233 [1] main/108/lua C> leaving orphan mode
imdb_1 | 2020-08-06 08:59:34.234 [1] main/108/lua I> set 'custom_proc_title' configuration option to "imdb@imdb-router"
imdb_1 | 2020-08-06 08:59:34.234 [1] main/111/checkpoint_daemon I> scheduled next checkpoint for Thu Aug 6 10:21:30 2020
imdb_1 | 2020-08-06 08:59:34.235 [1] main/108/lua I> set 'read_only' configuration option to true
imdb_1 | 2020-08-06 08:59:34.235 [1] main/108/lua I> Making sure user "admin" exists...
imdb_1 | 2020-08-06 08:59:34.235 [1] main/108/lua I> Setting password for user "admin" ...
imdb_1 | 2020-08-06 08:59:34.235 [1] main/108/lua I> set 'read_only' configuration option to false
imdb_1 | 2020-08-06 08:59:34.235 [1] wal/101/main xlog.c:1020 W> fallocate is not supported, proceeding without it
imdb_1 | 2020-08-06 08:59:34.236 [1] main/108/lua I> set 'read_only' configuration option to true
imdb_1 | 2020-08-06 08:59:34.236 [1] iproto/101/main I> binary: bound to 0.0.0.0:3301
imdb_1 | 2020-08-06 08:59:34.236 [1] main/107/remote_control/0.0.0.0:3301 utils.c:1018 E> LuajitError: builtin/socket.lua:83: attempt to use closed socket
imdb_1 | 2020-08-06 08:59:34.236 [1] main/108/lua I> set 'listen' configuration option to "3301"
DI
imdb_1 | 2020-08-06 08:59:34.236 [1] main/108/lua I> Instance state changed: RecoveringSnapshot -> ConnectingFullmesh
imdb_1 | 2020-08-06 08:59:34.236 [1] main/108/lua I> connecting to 1 replicas
imdb_1 | 2020-08-06 08:59:35.341 [1] main/103/lua I> Could not reach node: 172.31.0.2:3301 - suspect
imdb_1 | 2020-08-06 08:59:39.142 [1] main/103/lua I> Node timed out: 172.31.0.2:3301 - dead
imdb_1 | 2020-08-06 08:59:55.275 [1] main/122/applier/admin@172.31.0.2:3301 I> can't connect to master
imdb_1 | 2020-08-06 08:59:55.275 [1] main/122/applier/admin@172.31.0.2:3301 coio.cc:108 !> SystemError connect, called on fd 18, aka 192.168.0.2:52754: Connection refused
imdb_1 | 2020-08-06 08:59:55.275 [1] main/122/applier/admin@172.31.0.2:3301 I> will retry every 1.00 second
imdb_1 | 2020-08-06 09:00:04.236 [1] main/108/lua C> failed to connect to 1 out of 1 replicas
imdb_1 | 2020-08-06 09:00:04.237 [1] main/108/lua I> synchronizing with 1 replicas
imdb_1 | 2020-08-06 09:00:04.237 [1] main/108/lua C> failed to synchronize with 1 out of 1 replicas
imdb_1 | 2020-08-06 09:00:04.237 [1] main/108/lua C> entering orphan mode
imdb_1 | 2020-08-06 09:00:04.238 [1] main/108/lua I> set 'replication' configuration option to ["admin@172.31.0.2:3301"]
imdb_1 | 2020-08-06 09:00:04.238 [1] main/108/lua I> Instance state changed: ConnectingFullmesh -> ConnectingFullmesh
imdb_1 | 2020-08-06 09:00:04.238 [1] main/108/lua confapplier.lua:503 W> Replication setup failed, instance orphaned
DI
назначен ip 192.168.0.2 - и он сам себя найти не может
YD
хостнейм у этого контейнера есть? такой, по которому можно извне заходить. четкий понятный не меняющийся
DI
dns-имя есть imdb
YD
вот его и используйте в качестве advertise_uri
DI
теперь понял, спасибо
PL
https://habr.com/en/company/mailru/blog/513912/
by @racktear
Сегодня я немного расскажу о том, какие есть мысли по поводу фейловера в tarantool/cartridge. Сначала пару слов про то, что такое cartridge: это кусок lua-кода, который работает внутри tarantool и объединяет тарантулы друг с другом в один условный «кластер». Это происходит за счет двух вещей:
- каждый тарантул знает сетевые адреса всех других тарантулов;
- тарантулы регулярно «пингуют» друг друга через UDP, чтобы понять кто жив, а кто нет. Тут я намеренно немного упрощаю, алгоритм пинга сложнее чем просто request-response, но это для разбора не сильно принципиально. Если интересно — погуглите алгоритм SWIM.
Внутри кластера все обычно разделяется на тарантулы «с состоянием» (master/replica) и «без состояния» (router). Тарантулы «с состоянием» ответственны за хранение данных, а тарантулы «без состояния» — за маршрутизацию запросов.
...
by @racktear
Сегодня я немного расскажу о том, какие есть мысли по поводу фейловера в tarantool/cartridge. Сначала пару слов про то, что такое cartridge: это кусок lua-кода, который работает внутри tarantool и объединяет тарантулы друг с другом в один условный «кластер». Это происходит за счет двух вещей:
- каждый тарантул знает сетевые адреса всех других тарантулов;
- тарантулы регулярно «пингуют» друг друга через UDP, чтобы понять кто жив, а кто нет. Тут я намеренно немного упрощаю, алгоритм пинга сложнее чем просто request-response, но это для разбора не сильно принципиально. Если интересно — погуглите алгоритм SWIM.
Внутри кластера все обычно разделяется на тарантулы «с состоянием» (master/replica) и «без состояния» (router). Тарантулы «с состоянием» ответственны за хранение данных, а тарантулы «без состояния» — за маршрутизацию запросов.
...
KY
процессор транзакций работает в одном потоке, да
KY
у винила может быть куча потоков, WAL работает отдельно и Relay тоже все отдельные треды
AK
тх — системный, внутри кооперативная многозадачность с "зелеными" потоками