I've asked and talk about this on a few threads, but want to dial this in a bit more...
My situation...
We currently are running on ES 5.5.1 (Cluster Environment) on Ubuntu 16.04. Recently, we attempted to upgrade the OS of our servers to Ubuntu 18.04, but for some unknown reason, once the upgrade was complete, it completely removed our ES install on each server??? (Still have no idea why this happened)
We were able to roll back to our backups, and we are back up and running on 16.04 and ES 5.5.1 Our ES data is on a seperate drive (SAN Space) and that wasn't removed with the upgrade. So the upgrade didn't remove the data. With this said, we want to retry the upgrade but this time backing up our Elasticsearch.yml and java.options files on each server. And if ES is removed again with the upgrade, we will simply reinstall ES on each server.
Now, with that said,
Since ES is completely gone from our servers, can we just install the latest ES 7.x and overwrite the Elasticsearch.yml and Java.options files from our backups for each server? Like i said, the ES indexes are currently 5.5.1 indexes, so will the latest 7.x be able to fully utilize these 5.5.1 indexes? From my research with the upgrade guides, it appears it should work without having to reindex, but wanted to verify with the experts first.
The upgrade section of the reference manual has instructions on how to upgrade your cluster. Since you're starting at 5.5.1 the recommended path is a three-step process: 5.5.1 -> 5.6.16 -> 6.8.15 -> 7.12.1. It also says this:
Elasticsearch can read indices created in the previous major version. If you have indices created in 5.x or before, you must reindex or delete them before upgrading to 7.12.1.
OHHH, so I must first reinstall 5.5.1 then upgrade to 7.x. Ok, question… Will 5.5.1 run on Ubuntu 20.04 and is there an install guide if I need to reinstall 5.5.1?
also,
Instead of 5.5.1, if I have to reinstall ES, can i just go to ES 5.6? My indexes are in 5.5.1 so if I just reinstall 5.6 will my 5.5.1 indexes be able to be used?
Ah, I misunderstood and thought you still had the 5.5.1 cluster running. If it's all stopped then you can start from 6.8.15 which can still read indices written by 5.x. That is, unless you also have indices from 2.x, if you still have any of them then you'll need to reindex them in 5.x first.
No, i have no indexes from 2.x. All the indexes are from 5.5.1. So if reinstall 6.8.15 do i have to reindex all the indexes before later upgrading to the latest 7.x?
The config will likely be different between 5.5 and 6.8, and typically slightly different between node (different node name, maybe different roles and resource allocation if you're using specialised nodes)
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.