Improvement on basic_setup() of helpers of test/system (Clean up phase)

[aliaslanturk]

/kind feature

This issue proposes an improvement on basic_setup() function of helpers in its clean up phase.

Current flow of the function

basic_setup() function includes following lines to clean up unnecessary images before running tests:

# Clean up all images except those desired
    found_needed_image=
    run_podman images --all --format '{{.Repository}}:{{.Tag}} {{.ID}}'
    for line in "${lines[@]}"; do
        set $line
        if [ "$1" == "$PODMAN_TEST_IMAGE_FQN" ]; then
            found_needed_image=1
        else 
            echo "# setup(): removing stray images $1 $2" >&3
            run_podman rmi --force "$1" >/dev/null 2>&1 || true
            run_podman rmi --force "$2" >/dev/null 2>&1 || true
        fi  
    done

The for loop iterates to find the test image and remove any other image if present. Removing part uses {{.Repository}}:{{.Tag}} and {{.ID}} keywords to feed podman rmi. First key (Repository:Tag) works fine. However, ID key causes a problem if the test image has been tagged before.

What happens?

When a system test (XXX-*.bats) uses default basic_setup(), test passes if normal conditions are present.
The problem occurs if the test image ($IMAGE/$PODMAN_TEST_IMAGE_FQN) was tagged before as in:
podman tag $IMAGE same_image_with:new-tag
Following output can be observed as a result of the failure :

# setup(): removing stray images localhost/same_image_with:new-tag 4abe85964f5c
not ok 1 podman info - basic test
# (from function `is' in file helpers.bash, line 417,
#  in test file 005-info.bats, line 26)
#   `is "$output" ".*$expect" "output includes '$expect'"' failed

Root cause

As far as I can trace it, the problem is caused by run_podman rmi --force "$2" >/dev/null 2>&1 || true which removes the test image since run_podman rmi --force "$1" >/dev/null 2>&1 || true already removed the new_tag.

An improvement can be provided to avoid this case. I have not find a flow that reproduces the problem naturally, yet.

[aliaslanturk]

I would like to work on the issue. Due to some obligation, I cannot until early May. If the issue is to be solved, and assigned to me, I am more than happy to do so when I get back.

[rhatdan]

I have assigned it to you, I don't think anyone is going to grab it. Lots of other work to be done.
Thanks.

[aliaslanturk]

Removing run_podman rmi --force "$2" >/dev/null 2>&1 || true would work. However, I will commit that change after being assured whether it serves a specific purpose or not.

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[rhatdan]

@edsantiago PTAL

[edsantiago]

@aliaslanturk does #10008 address your concerns?

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[rhatdan]

Since we never heard back, I am going to assume yes, Please reopen if I am mistaken.