简体   繁体   English

在Solr 4.0中,从属节点是否可以立即看到向主节点提交的软提交文档?

[英]In Solr 4.0, are soft committed docs to master node visible to slave nodes immediately?

My search backend tries to upgrade to Solr 4.0 from Solr 3.6. 我的搜索后端尝试从Solr 3.6升级到Solr 4.0。

The existing arch based on Solr 3.6 has 1 master Solr node with 2 slave Solr nodes for load balancing, whose master node is for updating docs only and slave nodes are read-only. 基于Solr 3.6的现有架构具有1个主Solr节点和2个从属Solr节点以进行负载平衡,它们的主节点仅用于更新文档,而从属节点为只读。 Everything N minutes, docs are committed to master node then slave nodes poll indexed docs and install them, so these docs are visible. 在N分钟的所有时间中,将文档提交到主节点,然后从属节点轮询索引的文档并安装它们,因此这些文档是可见的。

In Solr 4.0, because soft committed docs are not sync-ed to disk files, my question is that if I upgrade to Solr 4.0 and don't use SolrCloud , are soft committed docs to master Solr node visible to slaves nodes immediately. 在Solr 4.0中,因为未将软提交的文档同步到磁盘文件,所以我的问题是,如果我升级到Solr 4.0并且不使用SolrCloud ,那么软提交的文档将成为可立即掌握从节点可见的Solr节点的软提交的文档。

If these are visible to slave, how it works, via zookeeper? 如果这些对奴隶可见,那么它通过Zookeeper如何工作?

The solr pull replication mechanism just copies the segment files on disk, nothing more. solr pull复制机制仅将片段文件复制到磁盘上,仅此而已。 Soft-committed documents are not on disk (not fsynced yet, which is done when a commit happens), that's why they won't be pulled while replicating. 软提交的文档不在磁盘上(尚未同步,在提交时完成),这就是为什么在复制时不会将其拉出的原因。

Like elasticsearch always did, SolrCloud uses the push approach. 像elasticsearch一样,SolrCloud使用推入方法。 When you index documents they will be indexed on all the nodes they are supposed to be indexed. 当您为文档建立索引时,它们将在应该被索引的所有节点上建立索引。 That's why soft-committed documents can be seen on all nodes they are indexed. 这就是为什么可以在所有软索引的文档的索引节点上看到它们的原因。

声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.

 
粤ICP备18138465号  © 2020-2024 STACKOOM.COM