Hello.
I have three VMs with elliptics, configured by file in attachment (ioserv.json, it differs for every node, of course in remote and address sections). I also installed on first (10.211.55.175) VM docker and pull latest reverbrain/backrunner image to run in container.
Then I run command:
docker run --rm -v /opt/backrunner:/mnt/backrunner reverbrain/backrunner /root/go/bin/bmeta -config /mnt/backrunner/etc/backrunner.json -bucket first
I see in log that bmeta hangs while trying to connect to one node, which is host for current docker container.
How I supposed to fix this in a correct and good way? I do not want to change docker network to host, because it's not a good way.
As far as I understand elliptics, removing this node from backrunner config will not change anything, because bmeta will receive full route table from elliptics server. Maybe I should point elliptics nodes to listen to both host and docker ips with route family "0"? Will it break other clients, because I suspect that docker host ip will be same for all nodes where I'll setup docker.
Why do I care? I'm trying to automate deployments and 15 seconds hang per node is not very good.
Why do I deploy backrunner on same host with elliptics? Because I don't have a lot of servers, also I have only 3 VMs in my test environment.
Regards,
Anatoly Popov.