南大通用GBase 8a镜像集群场景，少量高配机器做备份集群

发表于 作者 laozizhu

GBase 8a的镜像集群，一般建议采用相同的拓扑结构(相同的机器数量，相同的分布策略)。如果用于实时备份场景，也可以考虑用大容量磁盘，少量节点做备份用镜像。

测试环境

3节点集群。其中前两个节点是VC1，设置为p 1 每个节点1个主分片）；第三个节点是VC2，设置为 p2 d 0(每个节点2个主分片)。

两个VC做镜像的前提

  • 主分片数量相同，都是2个；不考虑副本
  • hashmap相同(vc2采用 initnodedatamap from vc1做的初始化)

gcadmin

[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$ gcadmin
CLUSTER STATE:         ACTIVE

=============================================================
|           GBASE COORDINATOR CLUSTER INFORMATION           |
=============================================================
|   NodeName   | IpAddress  | gcware | gcluster | DataState |
-------------------------------------------------------------
| coordinator1 | 10.0.2.101 |  OPEN  |   OPEN   |     0     |
-------------------------------------------------------------
=============================================
|     GBASE VIRTUAL CLUSTER INFORMATION     |
=============================================
|    VcName    | DistributionId |  comment  |
---------------------------------------------
|     vc1      |       1        | vc_101    |
---------------------------------------------
|     vc2      |       2        | vc_102    |
---------------------------------------------

2 virtual cluster: vc1, vc2
1 coordinator node
0 free data node

vc1的信息

2台机器，每个机器一个主分片。

[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$ gcadmin showcluster vc vc1
CLUSTER STATE:         ACTIVE
VIRTUAL CLUSTER MODE:  NORMAL

=============================================
|     GBASE VIRTUAL CLUSTER INFORMATION     |
=============================================
|    VcName    | DistributionId |  comment  |
---------------------------------------------
|     vc1      |       1        | vc_101    |
---------------------------------------------
=========================================================================================================
|                                 VIRTUAL CLUSTER DATA NODE INFORMATION                                 |
=========================================================================================================
| NodeName |                IpAddress                 | DistributionId | gnode | syncserver | DataState |
---------------------------------------------------------------------------------------------------------
|  node1   |                10.0.2.101                |       1        | OPEN  |    OPEN    |     0     |
---------------------------------------------------------------------------------------------------------
|  node2   |                10.0.2.102                |       1        | OPEN  |    OPEN    |     0     |
---------------------------------------------------------------------------------------------------------

2 data node

[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$ gcadmin showdistribution  vc vc1

                                 Distribution ID: 1 | State: new | Total segment num: 2

             Primary Segment Node IP                   Segment ID                 Duplicate Segment node IP
========================================================================================================================
|                   10.0.2.101                   |         1          |                   10.0.2.102                   |
------------------------------------------------------------------------------------------------------------------------
|                   10.0.2.102                   |         2          |                   10.0.2.101                   |
========================================================================================================================
[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$

VC2的信息

1台机器，有2个主分片。

[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$ gcadmin showcluster vc vc2
CLUSTER STATE:         ACTIVE
VIRTUAL CLUSTER MODE:  NORMAL

=============================================
|     GBASE VIRTUAL CLUSTER INFORMATION     |
=============================================
|    VcName    | DistributionId |  comment  |
---------------------------------------------
|     vc2      |       2        | vc_102    |
---------------------------------------------
=========================================================================================================
|                                 VIRTUAL CLUSTER DATA NODE INFORMATION                                 |
=========================================================================================================
| NodeName |                IpAddress                 | DistributionId | gnode | syncserver | DataState |
---------------------------------------------------------------------------------------------------------
|  node1   |                10.0.2.115                |       2        | OPEN  |    OPEN    |     0     |
---------------------------------------------------------------------------------------------------------

1 data node

[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$ gcadmin showdistribution  vc vc2

                                 Distribution ID: 2 | State: new | Total segment num: 2

             Primary Segment Node IP                   Segment ID                 Duplicate Segment node IP
========================================================================================================================
|                   10.0.2.115                   |         1          |                                                |
------------------------------------------------------------------------------------------------------------------------
|                   10.0.2.115                   |         2          |                                                |
========================================================================================================================
[gbase@gbase_rh7_001 gcinstall_9.5.2.45.5]$

创建镜像库

初始化并设置vc1到vc2的镜像。

gbase> use vc vc1;
Query OK, 0 rows affected (Elapsed: 00:00:00.00)

gbase> initnodedatamap;
Query OK, 1 row affected (Elapsed: 00:00:00.51)

gbase> use vc vc2;
Query OK, 0 rows affected (Elapsed: 00:00:00.00)

gbase> initnodedatamap from vc1;
Query OK, 1 row affected, 1 warning (Elapsed: 00:00:00.36)

gbase> show warnings;
+---------+------+----------------------------------------------------------+
| Level   | Code | Message                                                  |
+---------+------+----------------------------------------------------------+
| Warning | 1042 | It is risky that some suffix is not in the backup state. |
+---------+------+----------------------------------------------------------+
1 row in set (Elapsed: 00:00:00.00)

gbase> create database vc1.testdb;
Query OK, 1 row affected (Elapsed: 00:00:00.01)

gbase> create database vc2.testdb;
Query OK, 1 row affected (Elapsed: 00:00:00.00)

gbase> alter database vc1.testdb set default mirror=vc2;
Query OK, 1 row affected (Elapsed: 00:00:00.00)

gbase> show mirror databases;
+--------------------+-----+-----------+
| Database           | VC  | MIRROR_VC |
+--------------------+-----+-----------+
| information_schema |     |           |
| performance_schema |     |           |
| gbase              |     |           |
| gctmpdb            |     |           |
| gclusterdb         | vc2 |           |
| testdb             | vc2 | vc1       |
+--------------------+-----+-----------+
6 rows in set (Elapsed: 00:00:00.00)

测试镜像表

gbase> use vc vc1;
Query OK, 0 rows affected (Elapsed: 00:00:00.00)

gbase> use testdb;
Query OK, 0 rows affected (Elapsed: 00:00:00.00)

gbase> create table t1(id int,name varchar(100));
Query OK, 0 rows affected (Elapsed: 00:00:00.21)

gbase> insert into t1 values(1,'From vc1');
Query OK, 1 row affected (Elapsed: 00:00:00.13)

gbase> select * from vc1.testd.t1;
ERROR 1146 (42S02): Table 'vc1.testd.t1' doesn't exist
gbase> select * from vc1.testdb.t1;
+------+----------+
| id   | name     |
+------+----------+
|    1 | From vc1 |
+------+----------+
1 row in set (Elapsed: 00:00:00.01)

gbase> select * from vc2.testdb.t1;
+------+----------+
| id   | name     |
+------+----------+
|    1 | From vc1 |
+------+----------+
1 row in set (Elapsed: 00:00:00.01)

gbase> ^CAborted

总结

镜像集群，和主分片数量有关，双方必须有相同的主分片数；和hashmap有关，双方nodedatamap表里的data_distribution_id的数量必须相同，且内容必须相同。

经测试，当前版本对distribution的数量没有严格检查，如果只创建distribution,但没有initnodedatamap，此时不影响镜像。但如果做了initnodedatamap,则两面的nodedatamap将不一样，建镜像会报错。

可以用下面的SQL语句检查。看每个hashkey,nodeid组合的数量(group by hashkey,nodeid)做count(*)， 如果每个hashkey和nodeid。

其中的id是你镜像vc当前在用的distribution编号，可以在gcadmin showdistribution vc VCNAME 看到。

select distinct(res.r), count(res.r) from (select count(*) r from  gbase.nodedatamap t where t.data_distribution_id in (1,2) group by  t.hashkey, t.nodeid) res group by res.r;
+---+--------------+
| r | count(res.r) |
+---+--------------+
| 2 |        65536 |
+---+--------------+
1 row in set (Elapsed: 00:00:00.42)
Post Views: 200

相关文章:

  1. 南大通用GBase8a V862版本节点替换前的准备工作和注意事项
  2. 南大通用GBase 8a集群数据库用户密码长度限制
  3. 南大通用GBase 8a数据库加载时用户名、密码、路径等有特殊字符的处理
  4. 南大通用GBase 8a元数据表gbase.table_distribution
  5. 南大通用GBase 8a的注释comment 功能介绍，更新注释的方法，最大长度限制
  6. 南大通用GBase 8a通过集群日志查看执行计划和每个阶段的整体耗时和各个节点的耗时做性能排查
  7. 南大通用GBase 8a的rename命令改用户名和改表名
  8. 南大通用GBase 8a加载hadoop 测试2.6.10版本3节点集群安装
  9. 南大通用GBase 8a空值查询，任意一列为空NULL或某个值的数据行的方法
  10. 南大通用GBase 8a 通过物理文件操作实现表备份还原的方法