We had some discussion at the office about restarting a node or service when there is 1 node only issue.
Some people find that it is a bit awkwark to follow the step from "Full-cluster restart" documentation for only one node. Just to be clear here, we are not restarting all nodes. But following the step 1 ( Disable shard allocation.) and 2 ( Stop indexing and perform a synced flush) and restart ONLY the node with issue. Not all of them.
We are using version 7.13.4.
Is somebody can confirm that we have to follow these step, or not, and why ?
It's more a general question than a specific issue.. I'm the only ELK Admin and I was on vacation last week and somebody restart one node because a '/var/log/' disk space issue. That person didn't know about ELK that much. So, could we get Elasticsearch database issue if a node is restart without "Disable shard allocation" like the documentation suggest about a cluster restart ?
Really you should follow the rolling restart instructions since it's just a single node. But note that this is all optional, Elasticsearch will cope even if you just pull the power cord out on a node. The guidance in the docs just helps reduce the amount of extra work that Elasticsearch has to do to recover, but it'll recover eventually no matter how you shut the node down.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.