Elasticsearch upgrade from 5.5 to 7

Hi everybody,

We are using an old version of Elastic Stack (Elasticsearch, Logstash, Filebeat and Kibana) that we wanted to upgrade to lastest version (6.7) released yesterday (26/03). But we were wondering if we wouldn't wait for the version 7 except that we noticed this: Elasticsearch 7.0 can read indices created in version 6.0 or above. An Elasticsearch 7.0 node will not start in the presence of indices created in a version of Elasticsearch before 6.0.

As this is a little ambiguous (it doesn't explicitly says that Elasticsearch 7.0 won't be able to write on an ELasticsearch 6.x index for example and if reindexing for a version before 6.x to an Elasticsearch 7.0 won't be acceptable) my question is... what is the best solution to upgrade?

Will it be possible to mount a cluster of Elasticsearch 7 nodes and use remote cluster reindex to reindex our indices from version 5.5 to the Elasticsearch 7 node?

P.S: when will the Elasticsearch 7 be released?

This means that you can start a 7.0 cluster using a data dir that has been created with Elasticsearch 6.x. It also means that you will still be able to write into those indices using your new 7.0 cluster.

You can also reindex from a 6.x or 5.x or 2.x cluster to a 7.x cluster.

P.S: when will the Elasticsearch 7 be released?

We never announce dates.

1 Like

Thank you for your reply.
I ask because they also say: IMPORTANT: Reindex indices from Elasticsearch 5.x or before
Indices created in Elasticsearch 5.x or before will need to be reindexed with Elasticsearch 6.x in order to be readable by Elasticsearch 7.x.
So I'm still a little bit confused

Yes, you will be able to use reindex-from-remote to pull data from a 5.5 cluster.

The docs about being able to read indices of particular versions is referring to being able to read the index data as it is written on disk by older versions. Reindex-from-remote still has an old cluster running and reading the data from disk, so these concerns don't apply.

It's quite common to have a single cluster and to upgrade it in place, so you need the upgraded cluster to be able to read the old index data itself. However if you have the resources to start up a new cluster and the time to wait for reindex-from-remote to complete then you don't need to do this.

Thanks for your reply. It's more clear for me now. Thanks guys @dadoonet @DavidTurner for the time you took to explain

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