我可以更改现有Citus表的分配方法吗？ [英] Can I change the distribution method on an existing Citus table?

问题描述

在从MySQL迁移到Citus集群的过程中，我使用了 range 分发方法。迁移已完成，但是现在我想将分发方法更改为 hash 。

During a migration from MySQL into a Citus cluster, I used the range distribution method. The migration is complete, but now I'd like to change the distribution method to hash.

对于已有数据的表，是否可以将分布方法从 range 更改为 hash ？ / strong>

Is there a way to change the distribution method from range to hash for an existing table with data already in it?

我想出了以下过程，但不确定是否有效：

I came up with the following procedure, but am not sure it's valid:

1. 更新 pg_dist_shard minvalue 和 maxvalue 列$ c>更改所有分片的表


2. 从更新 pg_dist_partition 表的分片存储类型列r 到 h


3. COMMIT;

1. Update the minvalue and maxvalue columns of the pg_dist_shard table for all shards being changed
2. Update the shard storage type column of the pg_dist_partition table from r to h
3. COMMIT;

推荐答案

这是一个好问题。目前，Citus没有提供直接的方法来更改现有数据的分区类型。

That is a good question. Currently, Citus does not provide a direct way to change partition type of existing data.

在范围分区中，记录根据其分区列值和分片最小值放置在分片中。 / max值。如果记录x驻留在分片y中，则表示 y.minvalue< = x.partition_column< = y.maxvalue 。

In range partitioning, records are placed in shards according to their partition column value and shard min/max values. If a record x resides in shard y, then it means y.minvalue <= x.partition_column <= y.maxvalue.

在哈希分区中，对分区列进行哈希处理，并根据此哈希值路由记录。因此，您在 pg_dist_shard 中看到的最小值/最大值是哈希函数结果的边界值。在这种情况下 y.minvalue< = hash（x.partition_column）< = y.maxvalue 。

In hash partitioning, the partition column is hashed and records are routed according to this hashed value. Therefore, min/max values you see in pg_dist_shard are the boundary values for the result of the hash function. In this case y.minvalue <= hash(x.partition_column) <= y.maxvalue.

因此，进行您提到的更改将最终导致分配不正确。为了从范围分区切换到哈希分区，应该重新分配数据。为此，我建议将数据重新加载到一个空的哈希分区表中。

Therefore, doing the changes you have mentioned would end up with an incorrect distribution. In order to switch from range partition to hash partition, the data should be re-distributed. To do that, I suggest reloading the data to an empty hash-partitioned table.

有关更多信息，您可以参考使用分布式表和哈希分布的Citus文档部分。

For more information, you can refer to Working with Distributed Tables and Hash Distribution sections of Citus Documentation.

这篇关于我可以更改现有Citus表的分配方法吗？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！