存算分离集群扩容咨询

Viewed 2

将集群从3台扩容到4台, show backends看状态是正常的, 但是创建的新的Table的时候, Tablet还是会分配到原来3台老的节点, 新节点的Tablet一直是0

show backends;
+---------------+--------------+---------------+--------+----------+----------+--------------------+---------------------+---------------------+-------+----------------------+-----------+------------------+-------------------+---------------+---------------+---------+----------------+--------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------+-----------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+----------+----------+-----------+---------------------+--------------+
| BackendId     | Host         | HeartbeatPort | BePort | HttpPort | BrpcPort | ArrowFlightSqlPort | LastStartTime       | LastHeartbeat       | Alive | SystemDecommissioned | TabletNum | DataUsedCapacity | TrashUsedCapacity | AvailCapacity | TotalCapacity | UsedPct | MaxDiskUsedPct | RemoteUsedCapacity | Tag                                                                                                                                                                                                                                 | ErrMsg | Version                     | Status                                                                                                                                                                                                         | HeartbeatFailureCounter | NodeRole | CpuCores | Memory    | LiveSince           | RunningTasks |
+---------------+--------------+---------------+--------+----------+----------+--------------------+---------------------+---------------------+-------+----------------------+-----------+------------------+-------------------+---------------+---------------+---------+----------------+--------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------+-----------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+----------+----------+-----------+---------------------+--------------+
| 1769595750726 | 172.16.32.4  | 9050          | 9060   | 8040     | 8060     | -1                 | 2026-04-15 10:22:20 | 2026-04-16 21:35:39 | true  | false                | 3390624   | 0.000            | 0.000             | 1.000 B       | 0.000         | 0.00 %  | 0.00 %         | 0.000              | {"cloud_unique_id" : "1:123:PMt0ksGk", "compute_group_status" : "NORMAL", "private_endpoint" : "", "compute_group_name" : "default_compute_group", "location" : "default", "public_endpoint" : "", "compute_group_id" : "OLw0Ya9T"} |        | doris-4.0.3-rc03-e9096296b8 | {"lastStreamLoadTime":-1,"isQueryDisabled":false,"isLoadDisabled":false,"isActive":true,"isShutdown":false,"currentFragmentNum":8,"lastFragmentUpdateTime":1776346538711}                                      | 0                       | mix      | 112      | 468.42 GB | 2026-04-16 21:35:39 | 5            |
| 1769595750747 | 172.16.32.5  | 9050          | 9060   | 8040     | 8060     | -1                 | 2026-04-15 20:56:30 | 2026-04-16 21:35:39 | true  | false                | 3390655   | 0.000            | 0.000             | 1.000 B       | 0.000         | 0.00 %  | 0.00 %         | 0.000              | {"cloud_unique_id" : "1:123:F0NnNV_K", "compute_group_status" : "NORMAL", "private_endpoint" : "", "compute_group_name" : "default_compute_group", "location" : "default", "public_endpoint" : "", "compute_group_id" : "OLw0Ya9T"} |        | doris-4.0.3-rc03-e9096296b8 | {"lastStreamLoadTime":-1,"isQueryDisabled":false,"isLoadDisabled":false,"isActive":true,"isShutdown":false,"currentFragmentNum":2,"lastFragmentUpdateTime":1776346538423}                                      | 0                       | mix      | 112      | 468.42 GB | 2026-04-16 21:35:39 | 3            |
| 1769595750748 | 172.16.32.6  | 9050          | 9060   | 8040     | 8060     | -1                 | 2026-04-15 10:24:38 | 2026-04-16 21:35:39 | true  | false                | 3390625   | 0.000            | 0.000             | 1.000 B       | 0.000         | 0.00 %  | 0.00 %         | 0.000              | {"cloud_unique_id" : "1:123:glwKVRU6", "compute_group_status" : "NORMAL", "private_endpoint" : "", "compute_group_name" : "default_compute_group", "location" : "default", "public_endpoint" : "", "compute_group_id" : "OLw0Ya9T"} |        | doris-4.0.3-rc03-e9096296b8 | {"lastStreamLoadTime":-1,"isQueryDisabled":false,"isLoadDisabled":false,"isActive":true,"isShutdown":false,"currentFragmentNum":2,"lastFragmentUpdateTime":1776346538734}                                      | 0                       | mix      | 112      | 468.42 GB | 2026-04-16 21:35:39 | 3            |
| 1776158626799 | 172.16.32.10 | 9051          | 9060   | 8040     | 8060     | -1                 | 2026-04-16 20:53:41 | 2026-04-16 21:35:39 | true  | false                | 0         | 0.000            | 0.000             | 1.000 B       | 0.000         | 0.00 %  | 0.00 %         | 0.000              | {"cloud_unique_id" : "1:123:ZB0mm4zu", "compute_group_status" : "NORMAL", "private_endpoint" : "", "compute_group_name" : "default_compute_group", "location" : "default", "public_endpoint" : "", "compute_group_id" : "OLw0Ya9T"} |        | doris-4.0.3-rc03-e9096296b8 | {"lastSuccessReportTabletsTime":"N/A","lastStreamLoadTime":-1,"isQueryDisabled":false,"isLoadDisabled":false,"isActive":true,"isShutdown":false,"currentFragmentNum":1,"lastFragmentUpdateTime":1776346525030} | 0                       | mix      | 96       | 370.29 GB | 2026-04-16 21:35:39 | 1            |
+---------------+--------------+---------------+--------+----------+----------+--------------------+---------------------+---------------------+-------+----------------------+-----------+------------------+-------------------+---------------+---------------+---------+----------------+--------------------+-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+--------+-----------------------------+----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------+-------------------------+----------+----------+-----------+---------------------+--------------+
SELECT BackendId, Host, TabletNum FROM backends();
+---------------+--------------+-----------+
| BackendId     | Host         | TabletNum |
+---------------+--------------+-----------+
| 1769595750748 | 172.16.32.6  |   3390626 |
| 1769595750726 | 172.16.32.4  |   3390625 |
| 1769595750747 | 172.16.32.5  |   3390653 |
| 1776158626799 | 172.16.32.10 |         0 |
+---------------+--------------+-----------+
4 rows in set (0.02 sec)

这是什么原因呢

1 Answers

在存算分离架构下,BE 本地的数据都是缓存,所以增加 Compute Group 时不需要做笨重的数据迁移过程,新的 Compute Group 只需在查询时缓存预热即可;他们的数据都是在对象或者hdfs上的,你看到那三个有数据是本地的cache的数据