Can ES 8 read and use the indexes created in ES 6

Hello,

We have indices created in ES 6 (external SAN storage) and we are about to create the latest LTS ES cluster on ES 8. can the data / indexes created in ES6 be read and used with ES 8?

No. Elasticsearch only supports reading indices created in the previous major version. There is a commercial feature that provides limited access to older indices, but that is not available with the free Basic license.

can he upgrade from es6 to es7 and then to es8?

Your set of options here are (roughly):

  • Run with a paid enterprise license and use archive indices
  • Upgrade from ES6 to ES7, reindex the data into ES7 format, then upgrade to ES8
  • Create a new ES8 cluster, and use reindex from remote to migrate the data from the old cluster to the new cluster. ES8 can reindex directly from earlier versions, including ES6
  • Create 2 clusters, one running ES7.17 and another running ES8.x
    • Load old ES6 data into the ES7 cluster
    • Put new data into the ES8 cluster
    • Configure the ES7 cluster to be able to search the ES8 cluster as a CCS remote cluster
    • If you need to search the old data, you use the ES7 cluster
    • If you just need to search the new data you use the ES8 cluster and get the latest features.
    • If you need to search across both sets of data, you use the ES7 cluster with CSS (but are restricted to the feature set from ES7)
2 Likes

question:

Do I have to re-index the data one index at a time or can i grab all the indices and reindex them into my new environment?
Do i run the reindex from my ingest nodes on from my new cluster?

I have my new cluster stood up, but of course when I went to point to the data, the data nodes broke due to incompatibility. So I have both environments stood up.

Not familiar with the ReIndex methods, i was looking at the docs, but didn't see how I can reindex to the new IP (Cluster) from the old IP (Cluster)

OR, would it be best to build a temp 1 cluster machine with ES 7 and point it to the old cluster SAN Store and run the temp cluster and allow it to index all the data into the 7.x format and once complete, turn back on the new cluster and go from there?

I linked to the docs for reindexing from a remote cluster. Are there particularly parts that don't make sense, or a specific step you're stuck on?

So with reindexing do I need to do it manually index by index. We have a few hundred indexes.

Our data is mnt into an external san storage mount and not on the individual nodes. If I standup a new cluster 7.17 and point the data nodes to the external drive, wouldn’t that handle the data and so forth?

Thanks for all your help on this.

See Reindex from multiple sources.

Thanks for that answer.

Would I have to do a Reindex if…
Our data is mnt into an external san storage mount and not on the individual nodes. If I standup a new cluster 7.17 and point the data nodes to the external drive, wouldn’t that handle the data and so forth?

Thanks for all your help on this.

Also

Is reindexing a slow process?