ES Auto Updating its version (Downgrade and Upgrade)

[vaidabhishek (Abhishek Vaid)]

• We deployed a 5 Node cluster on Ubuntu 16.04 using dpkg. Strangely enough we've been facing an issue where ES updates the version. It's sometimes 6.2.2 going to 6.2.1 or the other way.

• Really don't know what is happening.

Vaid, Abhishek

[tomriley]

Hey,

Have you verified that all 5 nodes are running the same version of Elasticsearch. It may be that when you hit the cluster you are hitting a different nodes running different versions.

You should verify all nodes are running the same version by using the cat API to view all the nodes:
curl -XGET 'localhost:9200/_cat/nodes?v&pretty'

However, if all nodes are running the same version it is unlikely to be an issue with Elasticsearch. Elasticsearch does not have any sort of automated upgrade/downgrade orchestration functionality so its more likely to be related to the underlying infrastructure and how it is orchestrated.

How do you build the 5 nodes? Do you use any sort of configuration management such as Puppet or Chef?

Cheers,
Tom

[vaidabhishek (Abhishek Vaid)]

Hi Tom,

• Thanks for response. I talked to the team and it turned out that there were some automatic scripts which were managing the spot instances. When a node went down, another one came up and ES was installed using dpkg, which resulted in version mismatch.

• But it's useful to know that ES has no built in mechanism for Auto-Update. Perhaps an extremely useful feature would be to trigger an Auto-Update for all things Elastic natively supported by the stack. One command, which causes ES nodes to have a rollover update, all plugins to get updated, Logstash, Kibana, and filebeat can then detect this update and plan to self-update themselves. ELK stack has converged a lot, but still independent and manual updates of all components still remain little tedious.

• But anyway, thanks for reply. My problem is identified.

Vaid, Abhishek