tests/content-origins: Verify expected vendor for all RPM packages #1947
Conversation
| case "${ID}" in | ||
| fedora) vendor='Fedora Project';; | ||
| rhel|rhcos) vendor='Red Hat, Inc.';; | ||
| *) echo "Unknown operating system ID=${ID}; skipping this test"; exit 0;; |
There was a problem hiding this comment.
I wonder if we should actually error here if an unknown vendor pops up?
There was a problem hiding this comment.
You mean an unknown ID right? We could, but it seems unnecessarily painful for e.g. someone who is creating a *coreos derivative and wants to run our tests.
I didn't want to scope in looking at c9s, partially because based on active discussions it seems highly likely we may end up e.g. using a COPR or a SIG repo at some point, and we don't need to be strict.
There was a problem hiding this comment.
I think that's fair.. one counter is to make this run on distros of fcos and rhcos so this test won't even run on others?
still this is fine as is if we don't want to do that.
There was a problem hiding this comment.
The tricky thing is today kola makes distros=rhcos run on centos stream https://github.com/coreos/coreos-assembler/blob/32e6d9f6664cfd05703d217e04f2c6f517e35f70/mantle/cmd/kola/options.go#L312
| # Verify the RPM %{vendor} flag for everything installed matches what we expect. | ||
| # | ||
| # - platforms: qemu | ||
| # - This test should pass everywhere if it passes anywhere. |
There was a problem hiding this comment.
I know you have some opposition to this so I'll say:
optional: mention exclusive=false here
There was a problem hiding this comment.
Right, my rationale against the duplicate comments is covered in coreos/coreos-assembler#3060 and coreos/coreos-assembler#3059
cgwalters
force-pushed
the
vendor-verification
branch
from
08b12b7
to
6b3447d
Compare
I don't have a strong opinion here. The idea of naming it "content-origins" was to be a bit independent of RPM, even though 98.2% of our content comes from RPMs. (Perhaps in the future we somehow verify the systemd units we pull from src/config/overlay*?). But yes OTOH it does only look at rpms today. |
As we move to support multiple streams (particularly for C9S versus RHEL) we should ensure that we're not accidentally including content from CentOS in official RHEL builds, and vice versa. The RPM vendor field is useful for this. For now, this test just verifies Fedora and RHEL vendors, and I tested on both.
cgwalters
force-pushed
the
vendor-verification
branch
from
6b3447d
to
a68ce11
Compare
As we move to support multiple streams (particularly for C9S versus RHEL)
we should ensure that we're not accidentally including content
from CentOS in official RHEL builds, and vice versa. The RPM vendor
field is useful for this.
For now, this test just verifies Fedora and RHEL vendors, and I
tested on both.