Question about Rolling Upgrade

Hi, there,

We have a ES cluster with version 6.3.2. We have three separate Master nodes and two hot data nodes. We want to roll upgrade to latest version 6.5.1. We followed below instruction:

https://www.elastic.co/guide/en/elasticsearch/reference/current/rolling-upgrades.html

The purpose of rolling upgrade is to keep the ES up and running during the upgrade time. Here is what I did and the problem I got:

First, I brought down master node one by one. Whenever I brought down one master node, I will check the kibana and the data ingestion work, both are fine when I did the rolling upgrade master nodes.

Second, when I brought down the first hot data node, I can't access Kibana, it complains internal errors. Four minutes later, after I finished hot data node upgrades, immediately I can access Kibana.

Third, I brought down second hot data node, and this time both the kibana and data ingestion work fine.

However, after I finished the rolling upgrade, I checked the index that keeps running for the whole time of the upgrade, I found there is a gap on all the charts. the time matches with the first hot data node down time. So now I am not sure if the rolling upgrade can keep the index ingestion.

Please advise.

Thanks!

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.