34be2320-1881-4630-b704-7afdf6e17feb

hi @Martynas

I have you on deck from Slack on your migration

1. Get your schema

   curl -X GET "https://your-source-instance/v1/schema"

   -H "Content-Type: application/json" > schema.json

2. Start a backup

   https://weaviate.io/developers/weaviate/configuration/backups

3. Restore a backup assuming your name and number of nodes between source and target are identical.

hi @angelo how would we store the backup file on a railway weaviate instance?

You would have the backup file on your machine and then use the API to send it to your target instance

I see - would you think nodes refer to kubernetes nodes?

Nodes are a generic concept in the eyes of Weaviate, so if you are running a non-HA instance, you can assume you are running 1

we have 3 instances on gcp

and as far as I can see we can't scale replicas on volume services on railway

we have a concern for migration, as we are scalling we might need replicas for weaviate

Yep- I will need some time to see how we can mirror this

(almost 2 AM my time, but I can get to this first thing)

!remind me to figure out replicated weaviate in 8 hours

we have a concern for migration, as we are scalling we might need replicas for weaviate

This is a bigger concern for us right now, I bet we can do migration anyway.

my gut feeling is that we can do multiple instances and set up a reverse proxy

but we would need to sync those instances

What's the biggest weaviate cluster your customers have right now?

I am not sure, will need to test and read up on it

or if you had solved issues with replicated volumes for any customer

for any db service

We just ask users to set up follower DBs

and then pool connections

I think weaviate has leaderless architecture

we'd still need a way to tell weaviate to coordinate

If we end up with scalling issues, would railway in any way support us to scale weaviate?

Yes

We'd just give your instance generous limits