Test changes in `advanced_ci` branch

[sylwiaszunejko]

The purpose of this PR is to run CI with workflow file added in advanced_ci branch in the main repo as it uses pull_request_target and that's why the CI doesn't run new workflow file.

[dkropachev]

@sylwiaszunejko , failing due to the https://github.com/scylladb/scylla-cluster-tests/issues/8721. Also you need to pass image name to the jenkins job

[sylwiaszunejko]

Also you need to pass image name to the jenkins job

@dkropachev What do you mean? I pass SCT_STRESS_IMAGE.scylla-bench=scylladb/gocql-extended-ci:scylla-bench-advanced_ci_test in the extra_environment_variables parameter

[dkropachev]

Also you need to pass image name to the jenkins job

@dkropachev What do you mean? I pass SCT_STRESS_IMAGE.scylla-bench=scylladb/gocql-extended-ci:scylla-bench-advanced_ci_test in the extra_environment_variables parameter

Sorry, now I see it. It is better to have commit I'd in the image name, to avoid possible caching problems.

Another question, how do we clean these images up?

[sylwiaszunejko]

Another question, how do we clean these images up?

We do not have a mechanism for that yet

[sylwiaszunejko]

It is better to have commit I'd in the image name, to avoid possible caching problems.

But which commit if PR contains many of them? Maybe repo_name-branch_name would be good enough and then we have one image for one PR always, so we do not have that many images piling up

[dkropachev]

It is better to have commit I'd in the image name, to avoid possible caching problems.

But which commit if PR contains many of them?

Of course last commit id.

Maybe repo_name-branch_name would be good enough and then we have one image for one PR always, so we do not have that many images piling up

We still going to have them and still going to deal with them, these images are not going to go away by themselves. In this regard there is no difference between having 1 image per branch or having 5 images per branch, you still need a code to handle them.

You also going to have opportunity to go back and run test with image from prior version of them PR, in rare cases could be handy.

[sylwiaszunejko]

It is better to have commit I'd in the image name, to avoid possible caching problems.

But which commit if PR contains many of them?

Of course last commit id.

Maybe repo_name-branch_name would be good enough and then we have one image for one PR always, so we do not have that many images piling up

We still going to have them and still going to deal with them, these images are not going to go away by themselves. In this regard there is no difference between having 1 image per branch or having 5 images per branch, you still need a code to handle them.

You also going to have opportunity to go back and run test with image from prior version of them PR, in rare cases could be handy.

OK, I pushed the change, and I am starting my work on an workflow that will periodically scan docker hub and remove old images

[sylwiaszunejko]

@dkropachev Could you look at it? The last two runs https://jenkins.scylladb.com/view/scylla-drivers/job/scylla-drivers/job/gocql/job/extended-ci/job/longevity-large-partition-asymmetric-cluster-3h-test/ were successful but one job in github action was canceled