Fixing issue with validate_ci_input_variables CI script
As one can see from this pipeline: https://gitlab.cern.ch/cta/CTA/-/pipelines/12496255
the validate_ci_input_variables script did not succeed validating correct EOS tag
ERROR: Image tag '9a1939e5.el9' not found in repository 'gitlab-registry.cern.ch/dss/eos/eos-ci'
despite the tag being in the repository:
guenther@macbookpro ~ % skopeo list-tags docker://gitlab-registry.cern.ch/dss/eos/eos-ci | grep 9a1939e5
"9a1939e5.el9",Following the script and trying to reproduce this error locally failed !
There might be a mismatch in behavior between the manual terminal and the Python run_cmd wrapper. It works when trying to run the command manually - see below. In Python with check=True, even a non-error warning printed to stderr can trigger a CalledProcessError, depending on how podman sets exit codes in scripts. We can remove the check and check error codes explicitly, or remove the warnings by adding the --log-level=error flag to the command. Lets try the latter first.
[cirunner@ctadev12 ~]$ podman manifest inspect gitlab-registry.cern.ch/dss/eos/eos-ci:9a1939e5.el9
WARN[0000] The manifest type application/vnd.docker.distribution.manifest.v2+json is not a manifest list but a single image.
{
"schemaVersion": 2,
"mediaType": "application/vnd.docker.distribution.manifest.v2+json",
"config": {
"mediaType": "application/vnd.docker.container.image.v1+json",
"size": 4786,
"digest": "sha256:6f338da52558b988f6334175948f6cd2b068c202ab9ba95204603f33aab7b644"
},
[...]