跳转至

Storage负载均衡Graph

用户可以使用BALANCE语句平衡分片和Raft leader的分布，或者删除冗余的Storage服务器。

前提条件Graph

为了平衡分片和Raft leader的分布，NebulaGraph中图空间的副本数都必须大于1。

均衡分片分布Graph

BALANCE DATA语句会开始一个任务，将NebulaGraph集群中的分片平均分配到所有Storage服务器。通过创建和执行一组子任务来迁移数据和均衡分片分布。

Danger

不要停止集群中的任何机器或改变机器的IP地址，直到所有子任务完成，否则后续子任务会失败。

示例Graph

以横向扩容NebulaGraph为例，集群中增加新的Storage服务器后，新服务器上没有分片。

  1. 执行命令Graph检查分片的分布。

    nebual> SHOW HOSTS;
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| Host        | Port | Status   | Leader count | Leader distribution               | Partition distribution |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged0" | 9779 | "ONLINE" | 4            | "basketballplayer:4"              | "basketballplayer:15"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged1" | 9779 | "ONLINE" | 8            | "basketballplayer:8"              | "basketballplayer:15"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged2" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:15"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged3" | 9779 | "ONLINE" | 0            | "No valid partition"              | "No valid partition"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged4" | 9779 | "ONLINE" | 0            | "No valid partition"              | "No valid partition"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "Total"     |      |          | 15           | "basketballplayer:15"             | "basketballplayer:45"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+

  2. 执行命令BALANCE DATA将所有分片均衡分布。

    nebula> BALANCE DATA;
+------------+
| ID         |
+------------+
| 1614237867 |
+------------+

  3. 根据返回的任务ID，执行命令BALANCE DATA <balance_id>检查任务状态。

    nebula> BALANCE DATA 1614237867;
+--------------------------------------------------------------+-------------------+
| balanceId, spaceId:partId, src->dst                          | status            |
+--------------------------------------------------------------+-------------------+
| "[1614237867, 11:1, storaged1:9779->storaged3:9779]"         | "SUCCEEDED"       |
+--------------------------------------------------------------+-------------------+
| "[1614237867, 11:1, storaged2:9779->storaged4:9779]"         | "SUCCEEDED"       |
+--------------------------------------------------------------+-------------------+
| "[1614237867, 11:2, storaged1:9779->storaged3:9779]"         | "SUCCEEDED"       |
+--------------------------------------------------------------+-------------------+
...
+--------------------------------------------------------------+-------------------+
| "Total:22, Succeeded:22, Failed:0, In Progress:0, Invalid:0" | 100               |
+--------------------------------------------------------------+-------------------+

  4. 等待所有子任务完成，负载均衡进程结束，执行命令SHOW HOSTS确认分片已经均衡分布。

    Note

    BALANCE DATA不会均衡leader的分布。均衡leader请参见Graph。

    nebula> SHOW HOSTS;
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| Host        | Port | Status   | Leader count | Leader distribution               | Partition distribution |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged0" | 9779 | "ONLINE" | 4            | "basketballplayer:4"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged1" | 9779 | "ONLINE" | 8            | "basketballplayer:8"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged2" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged3" | 9779 | "ONLINE" | 0            | "No valid partition"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged4" | 9779 | "ONLINE" | 0            | "No valid partition"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "Total"     |      |          | 15           | "basketballplayer:15"             | "basketballplayer:45"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+

如果有子任务失败，请重新执行BALANCE DATA。如果重做负载均衡仍然不能解决问题，请到Graph寻求帮助。

停止负载均衡Graph

停止负载均衡，请执行命令BALANCE DATA STOP。

  • 如果没有正在执行的负载均衡任务，会返回错误。
  • 如果有正在执行的负载均衡任务，会返回停止的任务ID（balance_id）。

BALANCE DATA STOP不会停止正在执行的子任务，而是取消所有后续子任务。用户可以执行命令BALANCE DATA <balance_id>检查停止的任务状态。

一旦所有子任务都完成或停止，用户可以再次执行命令BALANCE DATA。

  • 如果前一个负载均衡任务的任何一个子任务失败，NebulaGraph会重新启动之前的负载均衡任务。
  • 如果前一个负载均衡任务的任何一个子任务都没有失败，NebulaGraph会启动一个新的的负载均衡任务。

移除Storage服务器Graph

移除指定的Storage服务器来缩小集群规模，可以使用命令BALANCE DATA REMOVE <host_list>。

示例Graph

如果需要移除以下两台Storage服务器。

服务器名称 IP地址 端口
storage3 192.168.0.8 9779
storage4 192.168.0.9 9779

请执行如下命令：

BALANCE DATA REMOVE 192.168.0.8:9779,192.168.0.9:9779;

NebulaGraph将启动一个负载均衡任务，迁移storage3和storage4中的分片，然后将服务器从集群中移除。

Note

已下线节点状态会显示为OFFLINE.

均衡leader分布Graph

BALANCE DATA只能均衡分片分布，不能均衡Raft leader分布。用户可以使用命令BALANCE LEADER均衡leader分布。

示例Graph

nebula> BALANCE LEADER;

用户可以执行SHOW HOSTS检查结果。

nebula> SHOW HOSTS;
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| Host        | Port | Status   | Leader count | Leader distribution               | Partition distribution |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged0" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged1" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged2" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged3" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "storaged4" | 9779 | "ONLINE" | 3            | "basketballplayer:3"              | "basketballplayer:9"   |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
| "Total"     |      |          | 15           | "basketballplayer:15"             | "basketballplayer:45"  |
+-------------+------+----------+--------------+-----------------------------------+------------------------+
最后更新: 2021年5月17日
Back to top