Closed jimbolimbo3 closed 3 years ago
172.17.0.x are IPs that a single Docker host assigns to containers. They are strictly internal to that Docker host; they aren't routable from the outside world. You can't connect to those IPs from other servers, at least not without either doing iptables configuration on both hosts or by using some kind of Docker mesh network. This doesn't have anything to do with XDCR, or even with Server for that matter.
Since you are exposing the ports with "docker run -p", all those ports will be available on the public IPs of the two servers themselves. So you should be able to hook up XDCR between the two servers using those IPs rather than the internal 172.17.0.x IPs.
Hi, I'm running two instances of Couchbase 7.0.0 beta on Docker in two different servers.
Since I noticed that multiple ports are needed for connection, I added all of them in the
docker run
comand:docker run -d --name db -p 9140:9140 -p 19130:19130 -p 21150:21150 -p 8091-8096:8091-8096 -p 11206-11211:11206-11211 -p 18091-18096:18091-18096 -p 21100:21100 -p 9119:9119 -p 9998-9999:9998-9999 -p 11213:11213 -p 21200:21200 -p 21300:21300 -p 21250:21250 -p 21350:21350 couchbase:7.0.0-beta
Evev I did that, I got these errors trying to use XDCR:
In one server (XDCR is bidirectional):
2021-04-06 10:12:23 172.17.0.3:genericPipeline.context.Start:Failed to start service CheckpointManager for bb121424227a24743e9fbc879d45ed2e/test/test. err=dial tcp 172.17.0.2:11210: connect: connection refused
In the other server:
2021-04-06 10:16:30 127.0.0.1:genericPipeline.context.Start:Failed to start service CheckpointManager for b07af2be29c3e94952a039c58aa12891/test/test. err=dial tcp 172.17.0.3:11210: connect: no route to host
Both server are Ubuntu server, updated and without any firewall on.
Is that a known issue? I had similar behaviour with stable release, by the way.