We have implemented AzureDevOps testing pipelines that test the deployment of our software, including the deployment of development environments. This is executed on an Azure VM using a remote powershell session.
On our previously installed version of Docker Desktop, 4.22, I was able able to start docker desktop on a remote powershell session by simply running the Docker Desktop executable and wait for it to startup.
We have now upgraded to version 4.37 and it no longer works correctly. The docker processes are starting, but attempts to query docker locks up.
For example, if you wait a few minutes and enter:
docker info
to get information on the status of docker, that command seems to lock up indefinitely. I had a test pipeline waiting for an hour until the pipeline locked up. I have replicated on a remote powershell session.
In version 4.37 I was excited to note the addition of new beta docker commands to automate the start/stop of docker desktop documented here:
However their is a note on “docker dekstop start” command that SSH is not supported. Its unclear if similar issues apply to winrm.
The docker desktop cli commands can act inconsistent when it gets in this strange state. “docker desktop status” will say docker is stopped. But complain that its already running if you use “docker desktop start” to start it. A bunch of docker processses are running.
You can login to a windows normally, run "docker desktop start " in that session, and use the docker info succesfully in a remote powershell session (effectively a seperate login to the same account)
With the addition of the new beta docker desktop cli commands its unclear if this is a supported use case or not. I can see it being yes or no. I’m trying to work out the appropiate strategy. Wait for the issue to get fixed, assuming it is one, or try to find an alternative.
A lot of ground. Any thoughts?