This is follow up on https://forums.docker.com/t/found-that-need-to-use-ucp-swarm-node-certs-on-ucp-controller-to-enable-multi-host-networking/5362/4?u=jojojojo1234
Long story short. In order to enable overlay networking in swarm, docker engine needs to start with following options:
--cluster-advertise xxxxxxxxx:12376 --cluster-store etcd://xxxxxxxxxxx:12379 --cluster-store-opt kv.cacertfile=/var/lib/docker/discovery_certs/ca.pem --cluster-store-opt kv.certfile=/var/lib/docker/discovery_certs/cert.pem --cluster-store-opt kv.keyfile=/var/lib/docker/discovery_certs/key.pem
This is in UCP 0.6.0 causing docker daemon to fail at start with error:
Registering as "xxxxxxxxxx:12376" in discovery failed: client: etcd cluster is unavailable or mis-configured.
The workaround is change startup option to use /var/lib/docker/ucp_discovery_certs, like this:
--cluster-advertise xxxxxxxxxx:12376 --cluster-store etcd://xxxxxxxxxxx:12379 --cluster-store-opt kv.cacertfile=/var/lib/docker/ucp_discovery_certs/ca.pem --cluster-store-opt kv.certfile=/var/lib/docker/ucp_discovery_certs/cert.pem --cluster-store-opt kv.keyfile=/var/lib/docker/ucp_discovery_certs/key.pem
folder /var/lib/docker/ucp_discovery_certs has to be manually created and all certs from /var/lib/docker/discovery_certs/ copied to new location.
I consider this to be bug in version 0.6.0
Note: UCP 0.5.0 used the old path /var/lib/docker/discovery_certs/ so this issue was not present there
This is follow up on https://forums.docker.com/t/found-that-need-to-use-ucp-swarm-node-certs-on-ucp-controller-to-enable-multi-host-networking/5362/4?u=jojojojo1234
Long story short. In order to enable overlay networking in swarm, docker engine needs to start with following options:
--cluster-advertise xxxxxxxxx:12376 --cluster-store etcd://xxxxxxxxxxx:12379 --cluster-store-opt kv.cacertfile=/var/lib/docker/discovery_certs/ca.pem --cluster-store-opt kv.certfile=/var/lib/docker/discovery_certs/cert.pem --cluster-store-opt kv.keyfile=/var/lib/docker/discovery_certs/key.pem
This is in UCP 0.6.0 causing docker daemon to fail at start with error:
Registering as "xxxxxxxxxx:12376" in discovery failed: client: etcd cluster is unavailable or mis-configured.
The workaround is change startup option to use /var/lib/docker/ucp_discovery_certs, like this:
--cluster-advertise xxxxxxxxxx:12376 --cluster-store etcd://xxxxxxxxxxx:12379 --cluster-store-opt kv.cacertfile=/var/lib/docker/ucp_discovery_certs/ca.pem --cluster-store-opt kv.certfile=/var/lib/docker/ucp_discovery_certs/cert.pem --cluster-store-opt kv.keyfile=/var/lib/docker/ucp_discovery_certs/key.pem
folder /var/lib/docker/ucp_discovery_certs has to be manually created and all certs from /var/lib/docker/discovery_certs/ copied to new location.
I consider this to be bug in version 0.6.0
Note: UCP 0.5.0 used the old path /var/lib/docker/discovery_certs/ so this issue was not present there