本文详细探讨了在Elasticsearch中设置集群时，节点配置（包括主节点、复制节点和节点ID）对数据分布的影响。特别关注了在节点重启后，数据在集群中的重新分配情况，并通过实例演示了数据段如何在服务恢复后被重新获取。同时，文章还深入讨论了在使用无复制配置时，如何实现数据在不同节点间的复制与迁移，以及通过API监控数据分布的方法。
cluster setting
shards : 5
replica : 0
There are 3 ES nodes.
NODE_1 master
NODE_2
NODE_3
After some indices are completed, using the segment api we can see that there are 5 primary active shards.
NODE_1 (master) shard1, shard2
NODE_2 shard3
NODE_3 shard4, shard5
Then master node service is shutdown(kill process directly), using the segment api we can see that there are 3 primary active shards.
NODE_2 (master) shard3
NODE_3 shard4, shard5
Then bring back the NODE_1, using the segment api we can see that there are 5 primary active shards.
NODE_1 shard1, shard2
NODE_2 (master) shard3
NODE_3 shard4, shard5
It means the index segments can be brought back after the ES service is back.
But with replica 0 , is it possible to copy the sharding to other nodes?
Even using the shutdown service rest api, the shard is not available after node is shutdown.
Wrapper service is necesary to use if you want to restart the ES service using rest api.
Every time a node is restarted, the node id changes.
扫一扫
669
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
