了解Gossip协议 [英] Understanding Gossip protocol

问题描述

我正在阅读akka群集文档，现在我正在 Gossip 部分。我不理解以下声明：

I'm reading akka cluster documentation and now I'm at the Gossip section. I did not understand the following statement:

使用Gossip协议传递集群成员资格，其中集群的
当前状态通过群集随机闲聊，
优先选择 未见最新版本的成员 。

Cluster membership is communicated using a Gossip Protocol, where the current state of the cluster is gossiped randomly through the cluster, with preference to members that have not seen the latest version.

很难想象。我有以下问题：

It is pretty hard to imagine. I have the following question:

问题：如果最新的更改仍然闲聊，节点如何知道哪些成员没有看到最新的更改。

Question: how does a node know which members have not seen the latest change if the latest change is still gossiped.

我的意思是，如果某个节点收到通知如何决定将其发送到何处？显然，它不应该发送回来，因为发件人看到了最新的变化。但其他成员呢？他们中的一些人可能已经看过状态，没有任何其他方法可以解决问题，而不是查询所有这些成员。但是如果我们查询一个节点，它会回答不，我还没有看到它在查询过程中有人可能会将状态发送给它...

I mean, if a node received a notification how it decide where to send it? Obviously, it should not send it back, as the sender have seen the latest change. But what about the other members? Some of them might have already seen the state and there is no any other way to figure it out than query all these members. But if we query one node and it will respond with "no, I have not seen it yet" during the querying someone might send the state to it...

推荐答案

在下一段中，本文讨论了使用向量时钟。它说

In the next paragraph the article talks about using vector clocks. It says

通过传递一组在八卦期间看到当前状态版本的节点来实现融合。

Convergence is implemented by passing a set of nodes that have seen current state version during gossip.

请注意，我之前从未在八卦协议中看到过这种情况，因为它通常会很快收敛。我将继续寻找源代码，但我记得有一些关于1000个节点在大约10-15步中收敛的情况，当时gossipers交换信息而不是仅仅推送信息。

Note that I have never seen this before in a gossip protocol because it typically converges pretty quickly on its own. I'll go looking for the source, but I recall something about 1000 nodes converging in roughly 10-15 steps when gossipers exchange information instead of just pushing information.

这篇关于了解Gossip协议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！