test/ctr.bats: fixes and improvements

[kolyshkin]

What type of PR is this?

/kind bug

What this PR does / why we need it:

• fix a few bugs found in ctr.bats tests.
• make tests more compact (+394 −1,180 lines) and readable (mostly by removing excessive run use).
• fix the code to pass shellcheck.
• introduces edit_json helper to modify json using jq
  (the plan is to switch code using sed or python to this one).

The test fixed are:

1. "annotations passed through".

Commit 5f49b2c added this test case.

Obviously, run crictl do not output anything to stdout,
and run grep do not fail if grep exits with non-zero exit code.
So the test was not testing anything.

2. "privileged ctr -- check for rw mounts".

It was broken from the very beginning, since it was supposed to run a
privileged container and check that /sys/fs/cgroup is mounted rw.

Commit d3a50d8 ("privileged: set mounts to rw") added
test/testdata/container_config_privileged.json file, but it is not
used by the test (or any other tests). It also checks for ro mount
while it should check for rw (based on what I understood). Finally,
it used incorrect pod config while creating a container (not sure
if it makes any difference, just observing).

The test case was later amended to not check ro/rw bits since it was
failing on some configurations (see commit 78badc8).

This commit fixes the config used for container (and adds edit_json
function to which hopefully other tests will be eventually converted),
fixes the check for rw mount, and fixes the pod config used to create
container.

While at it, remove the never used container_config_privileged.json.

3. "ctr lifecycle", "ctr execsync should not overwrite initial spec args", "privileged ctr -- check for rw mounts".

Commits a2ec1d4 and 247d465 added a few checks using jq.
Alas, those checks do not work (and never worked); jq always succeeds.

This happened because

1. in run cmd1 ... | cmd2 the part starting with the pipe
   character is not part of run statement;

2. run eats cmd1 ... output (into $output variable);

so cmd2 is provided with empty input.

Now,

3. jq with empty input does not run any filters and thus succeeds
   (even with -e, see [1]).

The fix is to add a separate check that the output is not empty.

While at it, remove run where it's not needed from the other places
in those three tests we fix.

[1] jqlang/jq#1628

Please see individual commits for more details.

Which issue(s) this PR fixes:

Fixes: #3662

Special notes for your reviewer:

Please review commit-by-commit.

Does this PR introduce a user-facing change?

None

[openshift-ci-robot]

Skipping CI for Draft Pull Request.
If you want CI signal for your change, please convert it to an actual PR.
You can still manually trigger a test run with /test all

[codecov]

Codecov Report

Merging #4253 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #4253   +/-   ##
=======================================
  Coverage   38.59%   38.59%           
=======================================
  Files         111      111           
  Lines        8893     8893           
=======================================
  Hits         3432     3432           
  Misses       5077     5077           
  Partials      384      384           

[openshift-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: kolyshkin, mrunalp

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [mrunalp]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[kolyshkin]

Not sure what is going on. I have amended and pushed this branch, but github is not picking it up here.

[kolyshkin]

/retest

[mrunalp]

/lgtm

[mrunalp]

/retest

[kolyshkin]

single failure with e2e_features_fedora (which I've seen before):

1/5226 Tests Failed.
expand_less
Kubernetes e2e suite: [sig-network] IngressClass [Feature:Ingress] should not set default value if no default IngressClass [Serial] expand_less	0s
/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/network/ingressclass.go:61
Oct 12 01:36:32.373: Unexpected error:
    <*errors.StatusError | 0xc003a5f9a0>: {
        ErrStatus: {
            TypeMeta: {Kind: "", APIVersion: ""},
            ListMeta: {
                SelfLink: "",
                ResourceVersion: "",
                Continue: "",
                RemainingItemCount: nil,
            },
            Status: "Failure",
            Message: "ingressclasses.networking.k8s.io \"ingressclass1\" already exists",
            Reason: "AlreadyExists",
            Details: {
                Name: "ingressclass1",
                Group: "networking.k8s.io",
                Kind: "ingressclasses",
                UID: "",
                Causes: nil,
                RetryAfterSeconds: 0,
            },
            Code: 409,
        },
    }
    ingressclasses.networking.k8s.io "ingressclass1" already exists
occurred
/go/src/k8s.io/kubernetes/_output/local/go/src/k8s.io/kubernetes/test/e2e/network/ingressclass.go:63

There is an open issue (kubernetes/kubernetes#92147) and looks like the related fix (kubernetes/kubernetes#93427 is in our codebase but it's not enough).

[kolyshkin]

/retest

[kolyshkin]

/retest

[kolyshkin]

No logs for "failed" integration_crun, it seems it just went astray.

/retest