Upgrading my application from ES 5.3 to 7.7

[taichi]

I'm going to ES of my application (it's large) from version 5.3 to 7.7
I have to fix my application codes, finding difference between the two versions.

Elastic documentation has many Breaking changes pages, corresponding to every minor upgrades (not only every major upgrades).

https://www.elastic.co/guide/en/elasticsearch/reference/5.6/breaking-changes-5.6.html
https://www.elastic.co/guide/en/elasticsearch/reference/6.8/breaking-changes-6.0.html
https://www.elastic.co/guide/en/elasticsearch/reference/7.x/breaking-changes-7.3.html

So do I have to check all the breaking changes pages (5.4, 5.5, ... 7.6, 7.7) and find where to fix in my application?

Or is there more easy way to find all the differences between 5.3 and 7.7?

[Steve_Mushero (ELKman.io (ELK Stack Management Tool))]

Look at deprecation log, an amazing wonder of Elasticsearch - it tells you if you are using things that will go away BUT from v5 to v7 it may not look are enough ahead, so you probably have to fix stuff and get to V6 (ideally 6.8, which has more pre-7 checks) to look at things that need to be fixed to V7.

You have to go 5->6->7 anyway so that log should help a lot. Also, yes, the breaking changes notes should be reviewed for the areas you use; you might also build a small or tiny test cluster to run the app against for various API checks/tests.

[taichi]

Thank you! It seems there is Deprecations API (https://www.elastic.co/guide/en/elasticsearch/reference/7.8/migration-api-deprecation.html), but it is not available in 5.x ...

[taichi]

Is deprecation logging available in version 5.x?
I could not find its documentation in 5.x
https://www.elastic.co/guide/en/elasticsearch/reference/current/logging.html#deprecation-logging

[warkolm (Mark Walkom)]

You might be better off running up a new cluster, doing a remote reindex to import the data, point your code to the new cluster and run some tests.

[Steve_Mushero (ELKman.io (ELK Stack Management Tool))]

Seems not available in 5.x; thought it was. Then as Mark says, build test cluster to run against, and read the docs - hopefully your use is simple; biggest thing often the doc types going away.