Hi ! We have a self-hosted Typesense on a single server running without issue. We are trying to move to a 3 nodes cluster but we encounter a lot of 408 errors on documents update. We have tried to increase typesense python client

connection_timeout_seconds

to 60 seconds but it didn't fix it. The server and cluster config are the same except

--reset-peers-on-error

enabled on cluster. Do you have any advice or solution to fix those errors ?

Sounds like the cluster was not established properly.

Thanks for your reply. Everything looks good, we have 1, 4 and 4.

Could you post the last 100 lines of logs from each node here?

After digging into nodes logs, it seems to come from our snapshot restoration. We get this message:

Unable to trigger snapshot as one or more of the peers reported unhealthy

. Read operations are working correctly, but write operations trigger 408. To restore the snapshot from our single server, we restore it to the 3 nodes. Is it the way to go ? NB: Typesense is running into docker