MariaDB Galera 集群设置问题

Posted

技术标签:

【中文标题】MariaDB Galera 集群设置问题【英文标题】:MariaDB Galera Cluster set up problems 【发布时间】:2014-07-30 09:31:34 【问题描述】:

我正在尝试启动并运行 mariadb 集群,但它不适合我。现在我在 64 位红帽 ES6 机器上使用 MariaDB Galera 5.5.36。我在这里通过这个 repo 安装了 mariadb:

[mariadb]
name = MariaDB
baseurl = http://yum.mariadb.org/5.5-galera/rhel6-amd64/
gpgkey=https://yum.mariadb.org/RPM-GPG-KEY-MariaDB
gpgcheck=1

在 server.conf 中,我在服务器 1 中有以下内容:

[mariadb]
log_error=/var/log/mariadb.log
query_cache_size=0
query_cache_type=0
binlog_format=ROW
default_storage_engine=innodb
innodb_autoinc_lock_mode=2
wsrep_provider=/usr/lib64/galera/libgalera_smm.so
wsrep_cluster_address=gcomm://192.168.211.133
wsrep_cluster_name='cluster'
wsrep_node_address='192.168.211.132'
wsrep_node_name='cluster1'
wsrep_sst_method=rsync

在服务器 2 上我有

[mariadb]
log_error=/var/log/mariadb.log
query_cache_size=0
query_cache_type=0
binlog_format=ROW
default_storage_engine=innodb
innodb_autoinc_lock_mode=2
wsrep_provider=/usr/lib64/galera/libgalera_smm.so
wsrep_cluster_address=gcomm://192.168.211.132
wsrep_cluster_name='cluster'
wsrep_node_address='192.168.211.133'
wsrep_node_name='cluster2'
wsrep_sst_method=rsync

当我使用以下命令启动服务器 1 时: sudo service mysql start --wsrep-new-cluster 它启动得很好,如果我打开 mysql 并检查 wsrep 的状态,它说一切都已启动并正在运行很好,但是当我尝试在第二台服务器上执行 sudo service mysql start 时,我在错误日志中得到以下信息:

140609 14:47:55 mysqld_safe Starting mysqld daemon with databases from /var/lib/mysql
140609 14:47:56 mysqld_safe WSREP: Running position recovery with --log_error='/var/lib/mysql/wsrep_recovery.i5qfm2' --pid-file='/var/lib/mysql/localhost.localdomain-recover.pid'
140609 14:47:57 mysqld_safe WSREP: Recovered position 85448d73-ebe8-11e3-9c20-fbc1995fee11:0
140609 14:47:57 [Note] WSREP: wsrep_start_position var submitted: '85448d73-ebe8-11e3-9c20-fbc1995fee11:0'
140609 14:47:57 [Note] WSREP: Read nil XID from storage engines, skipping position init
140609 14:47:57 [Note] WSREP: wsrep_load(): loading provider library '/usr/lib64/galera/libgalera_smm.so'
140609 14:47:57 [Note] WSREP: wsrep_load(): Galera 25.3.2(r170) by Codership Oy <info@codership.com> loaded successfully.
140609 14:47:57 [Note] WSREP: CRC-32C: using hardware acceleration.
140609 14:47:57 [Note] WSREP: Found saved state: 85448d73-ebe8-11e3-9c20-fbc1995fee11:-1
140609 14:47:57 [Note] WSREP: Passing config to GCS: base_host = 192.168.211.133; 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 = 128M; gcache.size = 128M; 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; repl.causal_read_timeout = PT30S; repl.commit_order = 3; repl.key_format = FLAT8; repl.proto_max = 5
140609 14:47:57 [Note] WSREP: Assign initial position for certification: 0, protocol version: -1
140609 14:47:57 [Note] WSREP: wsrep_sst_grab()
140609 14:47:57 [Note] WSREP: Start replication
140609 14:47:57 [Note] WSREP: Setting initial position to 85448d73-ebe8-11e3-9c20-fbc1995fee11:0
140609 14:47:57 [Note] WSREP: protonet asio version 0
140609 14:47:57 [Note] WSREP: Using CRC-32C (optimized) for message checksums.
140609 14:47:57 [Note] WSREP: backend: asio
140609 14:47:57 [Note] WSREP: GMCast version 0
140609 14:47:57 [Note] WSREP: (0c085f34-efe5-11e3-9f6b-8bfd1706e2a4, 'tcp://0.0.0.0:4567') listening at tcp://0.0.0.0:4567
140609 14:47:57 [Note] WSREP: (0c085f34-efe5-11e3-9f6b-8bfd1706e2a4, 'tcp://0.0.0.0:4567') multicast: , ttl: 1
140609 14:47:57 [Note] WSREP: EVS version 0
140609 14:47:57 [Note] WSREP: PC version 0
140609 14:47:57 [Note] WSREP: gcomm: connecting to group 'cluster', peer '192.168.211.132:,192.168.211.134:'
140609 14:48:00 [Warning] WSREP: no nodes coming from prim view, prim not possible
140609 14:48:00 [Note] WSREP: view(view_id(NON_PRIM,0c085f34-efe5-11e3-9f6b-8bfd1706e2a4,1) memb 
        0c085f34-efe5-11e3-9f6b-8bfd1706e2a4,0
 joined 
 left 
 partitioned 
)
140609 14:48:01 [Warning] WSREP: last inactive check more than PT1.5S ago (PT3.50775S), skipping check
140609 14:48:31 [Note] WSREP: view((empty))
140609 14:48:31 [ERROR] WSREP: failed to open gcomm backend connection: 110: failed to reach primary view: 110 (Connection timed out)
         at gcomm/src/pc.cpp:connect():141
140609 14:48:31 [ERROR] WSREP: gcs/src/gcs_core.c:gcs_core_open():196: Failed to open backend connection: -110 (Connection timed out)
140609 14:48:31 [ERROR] WSREP: gcs/src/gcs.c:gcs_open():1291: Failed to open channel 'cluster' at 'gcomm://192.168.211.132,192.168.211.134': -110 (Connection timed out)
140609 14:48:31 [ERROR] WSREP: gcs connect failed: Connection timed out
140609 14:48:31 [ERROR] WSREP: wsrep::connect() failed: 7
140609 14:48:31 [ERROR] Aborting

140609 14:48:31 [Note] WSREP: Service disconnected.
140609 14:48:32 [Note] WSREP: Some threads may fail to exit.
140609 14:48:32 [Note] /usr/sbin/mysqld: Shutdown complete

140609 14:48:32 mysqld_safe mysqld from pid file /var/lib/mysql/localhost.localdomain.pid ended

我不知道为什么第二台服务器无法检测到集群已启动并正在运行。这些机器可以很好地相互通信,我可以从一台机器SSH到另一台机器,它们可以相互ping通。我尝试删除 galera 缓存,尝试降级我的 mariadb galera 版本,尝试禁用 SELinux,尝试以其他用户身份运行 mysql 服务,验证正确的端口已打开,尝试在具有不同 IP 地址的不同计算机上的 2 个 VM 上运行它们等。有谁知道这里发生了什么,因为我一直在寻找 3 天试图解决这个问题,但似乎没有解决方案适合我。

【问题讨论】:

【参考方案1】:

我认为您需要在 wsrep_cluster_address 参数中列出所有 IP。

wsrep_cluster_address=gcomm://192.168.211.132,192.168.211.133

这应该在两台主机上完成。顺便说一句,您可能需要三个节点而不是两个节点,以避免出现脑裂情况。

【讨论】:

【参考方案2】:

这是我解决类似问题的方法。

CentOS 7 w/MariaDB Galera 10.1。

Node2 我看到了这个:

016-12-27 15:40:38 140703512762624 [Warning] WSREP: no nodes coming from prim view, prim not possible

做了一些阅读后,我尝试在 node1 上运行它。

service mysql start --wsrep-new-cluster

但这失败了,在日志中,我发现了这个......

2016-12-27 15:44:08 140438853814528 [ERROR] WSREP: It may not be safe to bootstrap the cluster from this node. It was not the last one to leave the cluster and may not contain all the updates. To force cluster bootstrap with this node, edit the grastate.dat file manually and set safe_to_bootstrap to 1 .

所以我编辑了文件/var/lib/mysql/grastate.dat,将safe_to_bootstrap 更改为1

然后我可以使用以下方法启动主节点:

service mysql start --wsrep-new-cluster

然后其他的,我就用了

service mysql start

注意:这是在演示预生产环境中。在通过同时重新启动所有服务器使一切正常工作后,我立即将其破坏:P,但我知道没有写入,并且数据库是同步的。如果您正在生产中并且发生这种情况,您可以使用以下内容来确定在哪个节点上运行“new-cluster”,这类似于说,让我成为主要的。

mysqld_safe --wsrep-recover

如果这是一个生产问题,我强烈建议阅读这篇文章并使用 CloneZilla 进行备份,然后再向损坏的客户端发送命令!

https://www.percona.com/blog/2014/09/01/galera-replication-how-to-recover-a-pxc-cluster/

【讨论】:

【参考方案3】:

集群必须在主节点上使用此命令启动:

galera_new_cluster

启动第一个节点后,即可成功启动集群中的其他节点。

【讨论】:

以上是关于MariaDB Galera 集群设置问题的主要内容,如果未能解决你的问题,请参考以下文章

无法引导 Galera 集群

程序猿成长日记 | MariaDB Galera Cluster多主集群设置技巧

Mariadb配置Galera集群

MariaDB Galera集群部署--技术流ken

mariadb galera启动问题

MariaDB GALERA 集群双节点部署