A quick hack to work around the current issue with lazydocker and SSH-based remote Docker context definitions.

I've used the excellent lazydocker before, but only occasionally, reverting to the docker CLI for most of my monitoring work. I thought I'd start using it again and learn more about it. While I can (and have) run lazydocker as a Docker container I wanted to run it locally in the Linux container of my Chromebook and connect remotely to my Docker engines running on various machines.

If you are going to run lazydocker as a Docker container then I recommend you build the image yourself (the Dockerfile is in the repo) as the image on DockerHub is rather out of date.

I manage these remote engines via Docker contexts. Running docker context ls here shows me this:

NAME         DESCRIPTION                               DOCKER ENDPOINT               
default      Current DOCKER_HOST based configuration   unix:///var/run/docker.sock   
docker *     Docker Host on PVE LXC                    ssh://dj@docker               
homeops      Docker Host on homeops                    ssh://dj@homeops              
kkhw42xrfy   M2 Air                                    ssh://user@kkhw42xrfy      
synology     Docker Host on Synology NAS               ssh://dj@synology

If you're interested in finding out how to define and use these contexts, see the post Remote access to Docker on my Synology NAS.

Currently there are some issues with lazydocker correctly using the SSH-based remote addresses - see for example issue 510 which, while closed, describes exactly what I experience even with the latest release, as does another user. I have no doubt that this will be sorted soon, but in the meantime I decided to set up a workaround, based on the fact that the DOCKER_HOST environment variable is correctly evaluated (see feat(docker): Honor the host specified in current docker context) even if the host info in the context is not.

If my context is set to docker, for example, like this:

docker context set docker

then if I start lazydocker I get an error:

Docker event stream returned error: error during connect: Get "http://dj%40docker/v1.25/events": dial tcp: lookup dj@docker: no such host

However, if I set DOCKER_HOST before the invocation, like this:

DOCKER_HOST=ssh://dj@docker lazydocker

then lazydocker connects successfully.

The context name docker here might be slightly confusing - it's because that's what I called the LXC container I set up on Proxmox for running a Docker engine.

I can use docker context inspect to look into the details of the context, which look like this:

[
    {
        "Name": "docker",
        "Metadata": {
            "Description": "Docker Host on PVE LXC"
        },
        "Endpoints": {
            "docker": {
                "Host": "ssh://dj@docker",
                "SkipTLSVerify": false
            }
        },
        "TLSMaterial": {},
        "Storage": {
            "MetadataPath": "/home/qmacro/.docker/contexts/meta/d548...",
            "TLSPath": "/home/qmacro/.docker/contexts/tls/d548..."
        }
    }
]

And using the --format option of docker context inspect like this:

docker context inspect --format '{{.Endpoints.docker.Host}}'

I can get the SSH-based host value ssh://dj@docker which I can then set in DOCKER_HOST and invoke lazydocker.

Putting this all together I have now an alias lado defined thus:

alias lado="DOCKER_HOST=$(docker context inspect --format '{{.Endpoints.docker.Host}}') lazydocker"

which sets the appropriate value, from the current context, in DOCKER_HOST, and invokes lazydocker.