Hi Team,
We would like to upgrade our Weaviate from docker.io/semitechnologies/weaviate:1.19.1 to docker.io/semitechnologies/weaviate:1.22.0 in our EKS environment. Could you please let us know if there will be any impact from this upgrade, or if we can directly change the image?
Thank you.
hi @mathieu !!
A good practice is to never skip release versions.
So:
There is a second approach, that is migrating your data and reindexing everything.
This will not have vectorization costs, as you will bring the vectors with you.
For this you spin up a Weaviate cluster on latest version (or use our hosted offerings) and use this guide to migrate the data over.
This may be the best approach as you can now compare the differences. But of course, it will depend on the scale of your operation.
Some things to consider on such a big jump in versions is that you shouldnât stop in 1.22. There were a lot of improvements since. You can find them here
in 1.23.7+ we introduced GRPC. So you need to expose this endpoint service to user newer versions of clients. This alone can bring you a 80% of improvement on batches and queries.
and in 1.25.+ we introduced RAFT, so if you are using multi node deployment, some actions are required
Let me know if this helps!
THanks!
@DudaNogueira Thank you for the solution. By the way, can we directly change the Weaviate image version, or do we need to create another Weaviate instance to migrate the data?
you can change the image tag.
The second instance is only if you want to migrate your data directly. For this second approach, you can jump from your current version to latest directly
Thanks!
@DudaNogueira Thank you, we have successfully upgraded it.
Thatâs awesome @Matthew_Leung !!
Thanks for choosing Weaviate 
