Hello, I’m testing XCDR with a setup on Amazon EC2 with 3 nodes. The XCDR functions as expected among the “cloud” nodes. I’m also testing whether I can replicate via my home network machines and the cloud instances. It fails, and I think it has to do with the cloud based Private IPs.
To test this, I turned off the firewall on my local and opened the cloud machines. This didn’t help and the error message is an I/O error with the Private IPs references. I’m curious as to why/ where it’s getting the private IPs? Also, I’m going from local => to cloud for replications.
I am setting up the Couchbase cluster on AWS. My Couchbase version is Enterprise Edition 5.1.1 build 5723
In each ec2 instance, I create my cluster with Docker containers. I am able to add nodes to a cluster [which each node IP is it docker IP like 172.18.0.2].
But I’m not able to create XCDR replication. Ideally, I have to fill IP/Hostname with my other instance IP, which Couchbase is running on it. However adding remote cluster is fine, but when I add the replication for a bucket I receive a couple of these errors:
TopoChangeDet:unexpected status code, 404, in _pre_replicate response. respMap=map[error:not_found reason:uuids_dont_match]
I double checked from the container I can curl :11210 which it means it can resolve it.
So to make the long story short, the problem arises because of resolving names. Somehow automatically the Public DNS or Public IP of instances are resolved to the private one, which is not interesting for us . Therefore the solution which is not nice, I totally replaced docker config with bare couchbase on the instances. In this case when u configure IP/Hostname of your cluster/servers with Public DNS , XCDR works fine