load-balancing - 使用sst :xtrabackup (galera)时节点加入集群的问题

Question

看来该节点正在加入集群，然后失败了……我尝试使用rsync和xtrabackup进行操作，并且在状态传输期间失败。在我看来，我似乎缺少了一些真正简单的东西，而且我无法对此付诸实践。任何帮助将不胜感激。
有关节点的更多信息
硕士-10.XXX.XXX.161
节点1-10.XXX.XXX.160
安装的软件包:
MariaDB兼容的MariaDB通用MariaDB的开发MariaDB共享的MariaDB客户端MariaDB的测试MariaDB的Galera服务器(v5.5.29-1)
galera(v23.2.4-1.rhel6)
percona-xtrabackup(v2.1.6-702.rhel6)
节点1的配置

[mysqld]
wsrep_cluster_address = gcomm://10.XXX.XXX.161
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_provider_options = gcache.size=4G; gcache.page_size=1G
wsrep_cluster_name = galera_cluster
default_storage_engine = InnoDB
innodb_autoinc_lock_mode = 2
innodb_locks_unsafe_for_binlog = 1
wsrep_sst_method = xtrabackup
wsrep_sst_auth = root:rootpassword
wsrep_node_name=1

主配置

[mysqld]
wsrep_cluster_address = gcomm://
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_provider_options = gcache.size=4G; gcache.page_size=1G
wsrep_cluster_name = galera_cluster
default_storage_engine = InnoDB
innodb_autoinc_lock_mode = 2
innodb_locks_unsafe_for_binlog = 1
wsrep_sst_method = rsync
wsrep_slave_threads = 4
wsrep_sst_auth = root:rootpassword
wsrep_node_name = 2

node1日志文件

131203 16:09:03 mysqld_safe Starting mysqld daemon with databases from /var/lib/mysql
131203 16:09:03 mysqld_safe WSREP: Running position recovery with --log_error=/tmp/tmp.f2EedjRjbQ
131203 16:09:08 mysqld_safe WSREP: Recovered position 359350ee-5c63-11e3-0800-6673d15135cd:2188
131203 16:09:08 [Note] WSREP: wsrep_start_position var submitted: '359350ee-5c63-11e3-0800-6673d15135cd:2188'
131203 16:09:08 [Note] WSREP: Read nil XID from storage engines, skipping position init
131203 16:09:08 [Note] WSREP: wsrep_load(): loading provider library '/usr/lib64/galera/libgalera_smm.so'
131203 16:09:08 [Note] WSREP: wsrep_load(): Galera 23.2.4(r147) by Codership Oy <info@codership.com]]> loaded succesfully.
131203 16:09:08 [Note] WSREP: Found saved state: 00000000-0000-0000-0000-000000000000:-1
131203 16:09:08 [Note] WSREP: Reusing existing '/var/lib/mysql//galera.cache'.
131203 16:09:08 [Note] WSREP: Passing config to GCS: base_host = 10.XXX.XXX.160; base_port = 4567; cert.log_conflicts = no; gcache.dir = /var/lib/mysql/; gcache.keep_pages_size = 0; gcache.mem_size = 0; gcache.name = /var/lib/mysql//galera.cache; gcache.page_size = 1G; gcache.size = 4G; gcs.fc_debug = 0; gcs.fc_factor = 1; gcs.fc_limit = 16; gcs.fc_master_slave = NO; gcs.max_packet_size = 64500; gcs.max_throttle = 0.25; gcs.recv_q_hard_limit = 9223372036854775807; gcs.recv_q_soft_limit = 0.25; gcs.sync_donor = NO; replicator.causal_read_timeout = PT30S; replicator.commit_order = 3
131203 16:09:08 [Note] WSREP: Assign initial position for certification: -1, protocol version: -1
131203 16:09:08 [Note] WSREP: wsrep_sst_grab()
131203 16:09:08 [Note] WSREP: Start replication
131203 16:09:08 [Note] WSREP: Setting initial position to 00000000-0000-0000-0000-000000000000:-1
131203 16:09:08 [Note] WSREP: protonet asio version 0
131203 16:09:08 [Note] WSREP: backend: asio
131203 16:09:08 [Note] WSREP: GMCast version 0
131203 16:09:08 [Note] WSREP: (8814b4ba-5c67-11e3-0800-91035d554a96, 'tcp://0.0.0.0:4567') listening at tcp://0.0.0.0:4567
131203 16:09:08 [Note] WSREP: (8814b4ba-5c67-11e3-0800-91035d554a96, 'tcp://0.0.0.0:4567') multicast: , ttl: 1
131203 16:09:08 [Note] WSREP: EVS version 0
131203 16:09:08 [Note] WSREP: PC version 0
131203 16:09:08 [Note] WSREP: gcomm: connecting to group 'galera_cluster', peer '10.XXX.XXX.161:'
131203 16:09:09 [Note] WSREP: declaring 7a9a87e8-5c67-11e3-0800-8cb6cba8f65a stable
131203 16:09:09 [Note] WSREP: Node 7a9a87e8-5c67-11e3-0800-8cb6cba8f65a state prim
131203 16:09:09 [Note] WSREP: view(view_id(PRIM,7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,2) memb {
     7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,
     8814b4ba-5c67-11e3-0800-91035d554a96,
} joined {
} left {
} partitioned {
})
131203 16:09:09 [Note] WSREP: gcomm: connected
131203 16:09:09 [Note] WSREP: Changing maximum packet size to 64500, resulting msg size: 32636
131203 16:09:09 [Note] WSREP: Shifting CLOSED -> OPEN (TO: 0)
131203 16:09:09 [Note] WSREP: Opened channel 'galera_cluster'
131203 16:09:09 [Note] WSREP: New COMPONENT: primary = yes, bootstrap = no, my_idx = 1, memb_num = 2
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: Waiting for state UUID.
131203 16:09:09 [Note] WSREP: Waiting for SST to complete.
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: sent state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: got state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515 from 0 (2)
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: got state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515 from 1 (1)
131203 16:09:09 [Note] WSREP: Quorum results:
     version    = 2,
     component  = PRIMARY,
     conf_id    = 1,
     members    = 1/2 (joined/total),
     act_id     = 2521,
     last_appl. = -1,
     protocols  = 0/4/2 (gcs/repl/appl),
     group UUID = 359350ee-5c63-11e3-0800-6673d15135cd
131203 16:09:09 [Note] WSREP: Flow-control interval: [23, 23]
131203 16:09:09 [Note] WSREP: Shifting OPEN -> PRIMARY (TO: 2521)
131203 16:09:09 [Note] WSREP: State transfer required:
     Group state: 359350ee-5c63-11e3-0800-6673d15135cd:2521
     Local state: 00000000-0000-0000-0000-000000000000:-1
131203 16:09:09 [Note] WSREP: New cluster view: global state: 359350ee-5c63-11e3-0800-6673d15135cd:2521, view# 2: Primary, number of nodes: 2, my index: 1, protocol version 2
131203 16:09:09 [Warning] WSREP: Gap in state sequence. Need state transfer.
131203 16:09:11 [Note] WSREP: Running: 'wsrep_sst_xtrabackup --role 'joiner' --address '10.XXX.XXX.160' --auth 'root:rootpassword' --datadir '/var/lib/mysql/' --defaults-file '/etc/my.cnf' --parent '13175''
131203 16:09:11 [Note] WSREP: Prepared SST request: xtrabackup|10.162.143.160:4444/xtrabackup_sst
131203 16:09:11 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:09:11 [Note] WSREP: Assign initial position for certification: 2521, protocol version: 2
131203 16:09:11 [Warning] WSREP: Failed to prepare for incremental state transfer: Local state UUID (00000000-0000-0000-0000-000000000000) does not match group state UUID (359350ee-5c63-11e3-0800-6673d15135cd): 1 (Operation not permitted)
      at galera/src/replicator_str.cpp:prepare_for_IST():442. IST will be unavailable.
131203 16:09:11 [Note] WSREP: Node 1 (1) requested state transfer from '*any*'. Selected 0 (2)(SYNCED) as donor.
131203 16:09:11 [Note] WSREP: Shifting PRIMARY -> JOINER (TO: 2525)
131203 16:09:11 [Note] WSREP: Requesting state transfer: success, donor: 0
tar: dbexport/db.opt: Cannot open: Permission denied
tar: Exiting with failure status due to previous errors
131203 16:10:22 [Note] WSREP: 0 (2): State transfer to 1 (1) complete.
131203 16:10:22 [Note] WSREP: Member 0 (2) synced with group.
WSREP_SST: [ERROR] Error while getting st data from donor node:  0, 2 (20131203 16:10:22.379)
131203 16:10:22 [ERROR] WSREP: Process completed with error: wsrep_sst_xtrabackup --role 'joiner' --address '10.XXX.XXX.160' --auth 'root:rootpassword' --datadir '/var/lib/mysql/' --defaults-file '/etc/my.cnf' --parent '13175': 32 (Broken pipe)
131203 16:10:22 [ERROR] WSREP: Failed to read uuid:seqno from joiner script.
131203 16:10:22 [ERROR] WSREP: SST failed: 32 (Broken pipe)
131203 16:10:22 [ERROR] Aborting
131203 16:10:24 [Note] WSREP: Closing send monitor...
131203 16:10:24 [Note] WSREP: Closed send monitor.
131203 16:10:24 [Note] WSREP: gcomm: terminating thread
131203 16:10:24 [Note] WSREP: gcomm: joining thread
131203 16:10:24 [Note] WSREP: gcomm: closing backend
131203 16:10:25 [Note] WSREP: view(view_id(NON_PRIM,7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,2) memb {
     8814b4ba-5c67-11e3-0800-91035d554a96,
} joined {
} left {
} partitioned {
     7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,
})
131203 16:10:25 [Note] WSREP: view((empty))
131203 16:10:25 [Note] WSREP: gcomm: closed
131203 16:10:25 [Note] WSREP: New COMPONENT: primary = no, bootstrap = no, my_idx = 0, memb_num = 1
131203 16:10:25 [Note] WSREP: Flow-control interval: [16, 16]
131203 16:10:25 [Note] WSREP: Received NON-PRIMARY.
131203 16:10:25 [Note] WSREP: Shifting JOINER -> OPEN (TO: 2607)
131203 16:10:25 [Note] WSREP: Received self-leave message.
131203 16:10:25 [Note] WSREP: Flow-control interval: [0, 0]
131203 16:10:25 [Note] WSREP: Received SELF-LEAVE. Closing connection.
131203 16:10:25 [Note] WSREP: Shifting OPEN -> CLOSED (TO: 2607)
131203 16:10:25 [Note] WSREP: RECV thread exiting 0: Success
131203 16:10:25 [Note] WSREP: recv_thread() joined.
131203 16:10:25 [Note] WSREP: Closing slave action queue.
131203 16:10:25 [Note] WSREP: Service disconnected.
131203 16:10:25 [Note] WSREP: rollbacker thread exiting
131203 16:10:26 [Note] WSREP: Some threads may fail to exit.
131203 16:10:26 [Note] /usr/sbin/mysqld: Shutdown complete
Error in my_thread_global_end(): 2 threads didn't exit
131203 16:10:31 mysqld_safe mysqld from pid file /var/run/mysqld/mysqld.pid ended

主日志文件

131203 16:08:47 [Warning] Recovery from master pos 103358630 and file mysql-bin.000131.
131203 16:08:47 [Note] Event Scheduler: Loaded 0 events
131203 16:08:47 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:08:47 [Note] WSREP: Assign initial position for certification: 2497, protocol version: 2
131203 16:08:47 [Note] /usr/sbin/mysqld: ready for connections.
Version: '5.5.29-MariaDB-log'  socket: '/var/lib/mysql/mysql.sock'  port: 3306  MariaDB Server, wsrep_23.7.3.rXXXX
131203 16:08:47 [Note] WSREP: Synchronized with group, ready for connections
131203 16:08:47 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:09:09 [Note] WSREP: declaring 8814b4ba-5c67-11e3-0800-91035d554a96 stable
131203 16:09:09 [Note] WSREP: Node 7a9a87e8-5c67-11e3-0800-8cb6cba8f65a state prim
131203 16:09:09 [Note] WSREP: view(view_id(PRIM,7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,2) memb {
     7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,
     8814b4ba-5c67-11e3-0800-91035d554a96,
} joined {
} left {
} partitioned {
})
131203 16:09:09 [Note] WSREP: New COMPONENT: primary = yes, bootstrap = no, my_idx = 0, memb_num = 2
131203 16:09:09 [Note] WSREP: STATE_EXCHANGE: sent state UUID: 8861cdd5-5c67-11e3-0800-cc70fcc5f515
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: sent state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: got state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515 from 0 (2)
131203 16:09:09 [Note] WSREP: STATE EXCHANGE: got state msg: 8861cdd5-5c67-11e3-0800-cc70fcc5f515 from 1 (1)
131203 16:09:09 [Note] WSREP: Quorum results:
     version    = 2,
     component  = PRIMARY,
     conf_id    = 1,
     members    = 1/2 (joined/total),
     act_id     = 2521,
     last_appl. = 2517,
     protocols  = 0/4/2 (gcs/repl/appl),
     group UUID = 359350ee-5c63-11e3-0800-6673d15135cd
131203 16:09:09 [Note] WSREP: Flow-control interval: [23, 23]
131203 16:09:09 [Note] WSREP: New cluster view: global state: 359350ee-5c63-11e3-0800-6673d15135cd:2521, view# 2: Primary, number of nodes: 2, my index: 0, protocol version 2
131203 16:09:09 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:09:09 [Note] WSREP: Assign initial position for certification: 2521, protocol version: 2
131203 16:09:11 [Note] WSREP: Node 1 (1) requested state transfer from '*any*'. Selected 0 (2)(SYNCED) as donor.
131203 16:09:11 [Note] WSREP: Shifting SYNCED -> DONOR/DESYNCED (TO: 2525)
131203 16:09:11 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:09:11 [Note] WSREP: Running: 'wsrep_sst_xtrabackup --role 'donor' --address '10.XXX.XXX.160:4444/xtrabackup_sst' --auth 'root:rootpassword' --socket '/var/lib/mysql/mysql.sock' --datadir '/var/lib/mysql/' --defaults-file '/etc/my.cnf' --gtid '359350ee-5c63-11e3-0800-6673d15135cd:2525''
131203 16:09:11 [Note] WSREP: sst_donor_thread signaled with 0
131203 16:10:20 [Note] WSREP: Provider paused at 359350ee-5c63-11e3-0800-6673d15135cd:2604
131203 16:10:22 [Note] WSREP: Provider resumed.
131203 16:10:22 [Note] WSREP: 0 (2): State transfer to 1 (1) complete.
131203 16:10:22 [Note] WSREP: Shifting DONOR/DESYNCED -> JOINED (TO: 2606)
131203 16:10:22 [Note] WSREP: Member 0 (2) synced with group.
131203 16:10:22 [Note] WSREP: Shifting JOINED -> SYNCED (TO: 2606)
131203 16:10:22 [Note] WSREP: Synchronized with group, ready for connections
131203 16:10:22 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:10:25 [Note] WSREP: Node 7a9a87e8-5c67-11e3-0800-8cb6cba8f65a state prim
131203 16:10:25 [Note] WSREP: view(view_id(PRIM,7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,3) memb {
     7a9a87e8-5c67-11e3-0800-8cb6cba8f65a,
} joined {
} left {
} partitioned {
     8814b4ba-5c67-11e3-0800-91035d554a96,
})
131203 16:10:25 [Note] WSREP: forgetting 8814b4ba-5c67-11e3-0800-91035d554a96 (tcp://10.XXX.XXX.160:4567)
131203 16:10:25 [Note] WSREP: New COMPONENT: primary = yes, bootstrap = no, my_idx = 0, memb_num = 1
131203 16:10:25 [Note] WSREP: STATE_EXCHANGE: sent state UUID: b5dda52e-5c67-11e3-0800-4b2360dd84f9
131203 16:10:25 [Note] WSREP: STATE EXCHANGE: sent state msg: b5dda52e-5c67-11e3-0800-4b2360dd84f9
131203 16:10:25 [Note] WSREP: STATE EXCHANGE: got state msg: b5dda52e-5c67-11e3-0800-4b2360dd84f9 from 0 (2)
131203 16:10:25 [Note] WSREP: Quorum results:
     version    = 2,
     component  = PRIMARY,
     conf_id    = 2,
     members    = 1/1 (joined/total),
     act_id     = 2607,
     last_appl. = 2597,
     protocols  = 0/4/2 (gcs/repl/appl),
     group UUID = 359350ee-5c63-11e3-0800-6673d15135cd
131203 16:10:25 [Note] WSREP: Flow-control interval: [16, 16]
131203 16:10:25 [Note] WSREP: New cluster view: global state: 359350ee-5c63-11e3-0800-6673d15135cd:2607, view# 3: Primary, number of nodes: 1, my index: 0, protocol version 2
131203 16:10:25 [Note] WSREP: wsrep_notify_cmd is not defined, skipping notification.
131203 16:10:25 [Note] WSREP: Assign initial position for certification: 2607, protocol version: 2
131203 16:10:30 [Note] WSREP:  cleaning up 8814b4ba-5c67-11e3-0800-91035d554a96 (tcp://10..XXX.XXX.160:4567)

Answer 1

问题在于，MariaDB的数据目录(可能是dbexport)中有一个数据库备份目录(/var/lib/mysql/)。在执行SST时，提供程序将扫描数据目录以查找要发送的文件。它看到了目录，并假定它是用于数据库的，因为那是data目录中目录的目的。删除备份目录解决了该问题。最佳做法是，不要更改/var/lib/中的任何内容；程序通常将其数据文件保存在其中，并且将其弄乱可能会导致类似的问题。

解决主要问题后，日志中会出现一条新消息:

[Warning] WSREP: Failed to prepare for incremental state transfer: Local state UUID (00000000-0000-0000-0000-000000000000) does not match group state UUID (359350ee-5c63-11e3-0800-6673d15135cd): 1 (Operation not permitted) at galera/src/replicator_str.cpp:prepare_for_IST():442. IST will be unavailable.

此消息是正常的。当节点加入galera群集时，它将尝试执行IST(增量状态传输)而不是完整的SST(状态快照传输)。如果节点以前是群集的一部分，并且节点离开时的状态与群集的当前状态之间的差异很小，则可以使用IST，它仅传输节点的当前状态与群集的状态之间的差异。这比传输所有数据快得多。如果该节点以前是群集的一部分，但很久以前就离开了，则它将需要执行SST。在这种情况下，加入节点的状态UUID为 00000000-0000-0000-0000-000000000000，这基本上意味着它对集群来说是新的。我运行了MariaDB/galera集群，并且每当IST不可用时，此消息就会使我感到烦恼。如果这不是警告，而是改写了，那就太好了。我不确定为什么里面有 Operation not permitted，但这没什么好担心的。

此外，建议您运行奇数个节点以防止出现脑裂情况。如果可能，您应该将另一个MariaDB服务器添加到集群，或者如果无法运行，请运行 garbd。 garbd充当群集中的节点，而不是数据库服务器。它使您可以拥有奇数个节点，而无需其他数据库服务器。