Cassandra：并发模式更改的解决 [英] Cassandra: Resolution of concurrent schema changes

问题描述

我无法找到Cassandra如何处理其模式的并发更改的信息。例如：

I have been unable to find information on how Cassandra handles concurrent changes to its schema. For example:

在Cassandra集群中，使用机器A，B，C，D，E

In Cassandra cluster with machines A,B,C,D,E

到机器A：

CREATE TABLE foo (
  id_column int,
  x_column1 varchar,
  x_column2 varchar,
  PRIMARY KEY (id_column)
)

机器B

CREATE TABLE foo (
  id_column int,
  y_another_column1 varchar,
  y_another_column2 varchar,
  PRIMARY KEY (id_column)
)

会发生什么？它失败了吗？最后的结果是两者的合并吗？如果是合并，如果已经选择了diff PKs列，那么它会失败吗？

What happens? Does it fail? Is the final result a merge of both? If it is a merge, would it fail if diff PKs columns had been chosen?

我猜测并发创建冲突模式是一种不寻常的情况。然而，发出并发ALTER TABLE语句需要类似的解析度，并且在现实世界的情况下更有可能。

I guess concurrent creation of conflicting schema is an unusual scenario. However issuing concurrent ALTER TABLE statements would need similar resolution and that more likely in a real world scenario.

推荐答案

2 JIRA ticket（已解决）提供有关实时模式更改和冲突解决的详细信息：

The following 2 posts and 2 JIRA tickets (resolved) provide details about live schema changes and conflict resolution:

• 实时模式更新（ nb ：版本适用于功能首次可用）


• Cassandra 1.1中的模式管理复兴。


• 允许并发模式迁移（1391）（固定为C * 1.1）


• 避免ID来自并发模式更改的冲突（3794）（在C * 1.2中固定）

• Live schema updates (nb: the version applies to when this feature was firstly available)
• The schema management renaissance in Cassandra 1.1.
• Allow concurrent schema migrations (1391) (fixed in C* 1.1)
• Avoid ID conflicts from concurrent schema changes (3794) (fixed in C* 1.2)

CASSANDRA-1391 中的评论，以获取所有详细信息。

You'll have to go through the comments in CASSANDRA-1391 to get all the details.

这篇关于Cassandra：并发模式更改的解决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！