Docker Community Forums

Share and learn in the Docker community.

Unable to upgrade DTR from 2.0.0 to 2.0.1


(Qadinc) #1

Unable to upgrade to latest version of DTR

I’m sitting at version 2.0.0 and I want to upgrade to 2.0.1. Not required, but I like the latest and greatest :slight_smile:

I get this error when doing running the upgrade (I’m using the self signed certificate and I also have --insecure-registry on the docker daemon running on the same machine as DTR and where I’m running the upgrade command. I’ve tested this on pulls from DTR. I pulled the cert and did a cat to make sure it comes up and it looks good.

Failed to get UCP CA: Get https://host.example.com:8443/ca: x509: certificate signed by unknown authority

For now we have to use the self generated cert.

Thanks for any suggestions


(Qadinc) #2

Just an update. I was able to get past the cert error when I installed a geotrust cert, but now I get this. I would think with an upgrade it would not check port usage since it is indeed using those ports.

ERRO[0020] Make sure that there is a node in your UCP cluster where port 80 and port 443 are open. Also confirm that all UCP node have the DTR images or can obtain them from Docker Hub.
ERRO[0020] Upgrade has failed. Try running it again.
FATA[0020] Problem running container ‘dtr-phase2’ from image ‘docker/dtr:2.0.3’: Couldn’t create container ‘dtr-phase2’ from image ‘docker/dtr:2.0.3’: Error response from daemon: Unable to find a node that satisfies the following conditions
[available container slots]
[container!=ucp-controller (soft=false)]
[node==]


(Patrick Devine) #3

You can get around the TLS error by either using the ‘–ucp-insecure-tls’ option, or alternatively you can do something like ‘$ curl --insecure https://:/ca > ucp-cert.pem’ and then use ‘–ucp-ca “$(cat ucp-cert.pem)”’ during your install.

We tried to cover this in https://docs.docker.com/docker-trusted-registry/install/ but it can be a little confusing.


(Patrick Devine) #4

What version of the docker engine are you using and is your DTR sharing the same node as your UCP controller? There is a bug with certain versions of the engine/UCP where the DTR containers won’t restart correctly if you reboot your system. The error message is a little cryptic, but it’s trying to prevent you into getting into that state.