激活用于计算任务的节点,但无数据存储功能。

huangapple go评论95阅读模式
英文:

Ignite nodes for computing task but without data storage

问题

我在多台网络服务器上使用Ignite,旨在实现内存中和持久化数据缓存的混合使用。

大多数节点参与任务计算和缓存访问,但只有部分节点参与将数据持久化到磁盘。目前,我通过NodeFilter来控制数据存储,该过滤器检查那些需要在磁盘上存储数据的服务器的属性。
然而,这仍然需要那些不需要在磁盘上存储数据的节点配置persistenceEnabled=true,这也要求这些节点有storagePath、walPath和walArchivePath,但实际上这些属性是不会被使用的。

是否有一种更简单的方法来配置节点以用于计算任务,但又不需要数据持久化存储呢?

英文:

I'm using Ignite for a mix of in-memory only and persistence data caches on several network servers.

Most nodes participate in task computing and cache access, but only some participate in data persistence on disk. Currently I'm controlling the data storage through NodeFilter that checks the attributes of those servers that shall store data on disk.<br>
However, this still requires the nodes that shall not store data on disk to configure persistenceEnabled=true, which also requires these nodes to have a storagePath, walPath and walArchivePath which wouldn't actually be used.

Is there an easier to way to configure nodes for computing tasks but without data persistence storage?

答案1

得分: 0

你可以将计算节点作为客户端节点启动,然后可以使用 ClusterGroup 在其上调度计算(默认情况下,计算仅发送到服务器节点)。\n

英文:

You can start compute nodes as client nodes, then you can schedule computations on them using ClusterGroup (by default, compute is only sent to server nodes).

huangapple
  • 本文由 发表于 2020年9月18日 19:59:10
  • 转载请务必保留本文链接:https://go.coder-hub.com/63955352.html
匿名

发表评论

匿名网友

:?: :razz: :sad: :evil: :!: :smile: :oops: :grin: :eek: :shock: :???: :cool: :lol: :mad: :twisted: :roll: :wink: :idea: :arrow: :neutral: :cry: :mrgreen:

确定