Technology Tales

Launching SAS Analytics Pro on Viya with automated Docker image clean-up

Published on 28^th November 2025 Estimated Reading Time: 5 minutes

For my freelancing, I have a licensed version of SAS Analytics Pro running in a Docker container on my main Linux workstation. Every time there is a new release, a new Docker is made available, which means that a few of them could accumulate on your system. Aside from taking up disk space that could have other uses, it also makes it tricky to automate the startup of the associate Docker container. Avoiding this means pruning the Docker images available on the system, something that also needs automation.

To make things clearer, let me work through the launch script that I use; this is called by the script that checks for and then downloads any new image that is available, should that be needed. First up is the shebang, and this uses the -e switch to exit the script in the event of there being an error. That puts a stop to any potentially destructive outcomes from later commands being executed afterwards and without having the input that they need.

#!/bin/bash -e

Next comes the command to shut down the existing container. Should a new image get instated, this would lock up the old one, preventing its removal. Also, doing the rest of the steps with an already running container will result in errors anyway.

if docker container ls -a --format '{{.Names}}' | grep -q '^sas-analytics-pro$'; then
    docker container stop sas-analytics-pro
fi

After that, the step to find the latest image is performed. Once, I did this by looping through the ages by days, weeks and months, hardly an elegant or robust approach. What follows is something all the more effective.

# Find latest SAS Analytics Pro image
IMAGE=$(docker image ls --format '{{.Repository}}:{{.Tag}} {{.CreatedAt}}' \
    | grep 'sas-analytics-pro' \
    | sort -k2,3r \
    | head -n 1 \
    | awk '{print $1}')

echo "Chosen image: $IMAGE"

Since there is quite a lot happening above, let us unpack the actions. The first part lists all Docker images, formatting each line to show the image name (repository:tag) followed by its creation timestamp: docker image ls --format '{{.Repository}}:{{.Tag}} {{.CreatedAt}}'. The next piece picks out all the images that are for SAS Analytics Pro: grep 'sas-analytics-pro'. The crucial step, sort -k2,3r, comes next and this sorts the results by the second and third fields (the creation date and time) in reverse order, so the newest images appear first. With that done, it is time to pick out the most recent image using head -n 1. To pick out the image name, you need awk '{print $1}. This wrapped within IMAGE=$(...) to assign the result to a variable that is printed to the console using an echo statement.

With the image selected, you can then spin up the container once you specify the other parameters to use and allow some sleep time afterwards before proceeding to the clean-up steps:

run_args="
-e SASLOCKDOWN=0
--name=sas-analytics-pro
--rm
--detach
--hostname sas-analytics-pro
--env RUN_MODE=developer
--env SASLICENSEFILE=[Path to SAS licence file]
--publish 8080:80
--volume ${PWD}/sasinside:/sasinside
--volume ${PWD}/sasdemo:/data2
--volume [location of SAS files on the system]:/data
--cap-add AUDIT_WRITE
--cap-add SYS_ADMIN
--publish 8222:22
"

if ! docker run -u root ${run_args} "$IMAGE" "$@" > /dev/null 2>&1; then
    echo "Failed to run the image."
    exit 1
fi

sleep 5

With the new container in action, the subsequent step is to find the older images and delete those. Again, the docker image command is invoked, with its output fed to a selection command for SAS Analytics Pro images. Once the current image has been removed from the listing by the grep -v command, the list of images to be deleted is assigned to the IMAGES_TO_REMOVE variable.

IMAGES_TO_REMOVE=$(docker image ls --format '{{.Repository}}:{{.Tag}}' \
    | grep 'sas-analytics-pro' \
    | grep -v "^$IMAGE$")

echo "Will remove older images:"
echo "$IMAGES_TO_REMOVE"

After that has happened, iterating through the list of images using a for loop will remove them one at a time using the docker image rm command:

for OLD in $IMAGES_TO_REMOVE; do
    echo "Removing $OLD"
    docker image rm "$OLD" || echo "Could not remove $OLD"
done

All this concludes the operation of spinning up a new SAS Analytics Pro Docker container while also removing any superseded Docker images. One last step is to capture the password to use for logging into the SAS Studio interface that is available at localhost:8080 or whatever address and port is being used to serve the application:

docker logs sas-analytics-pro 2>&1 | grep "Password=" > pw.txt

Folding updating and housekeeping into the same activity as spinning up the Docker container means that I need not think of doing anything else. The time taken by the other activities repay the effort by always having the latest version running in a tidy environment. That just saves having to remember to do all of this, which is what is needed without automation.