SDCICD-1358: custom task to get OLM bundle version #1255
Conversation
jbpratt
force-pushed
the
SDCICD-1358
branch
2 times, most recently
from
328bc5e
to
b01072e
Compare
|
/ok-to-test |
| | sed -n '/"type": "olm.package"/,/}/p' \ | ||
| | sed -n 's/.*"version": "\(.*\)".*/\1/p' \ | ||
| | tr -d '\n' \ | ||
| > "$(results.bundle-version.path)" |
There was a problem hiding this comment.
That seems rather fragile, e.g. it wouldn't work if the JSON was printed compactly or if the order of the keys was different
{
"value": {
"packageName": "osd-example-operator",
"version": "4.16.0-28f1940"
},
"type": "olm.package"
}Is there reason to believe that opm would always format the output in a way that will work with this sed command?
There was a problem hiding this comment.
We control the way opm is ran (with json output), and we aren't receiving arbitrarily formatted JSON, I would assume this to be pretty stable. I don't think there is a way to print the json compactly through opm render, the only other output option being YAML through the --output=yaml flag. I would have preferred to use JQ but I can't find an image with both opm and jq in it. Your suggestion below for using JQ is a great improvement I can follow up with.
The order of the keys shouldn't matter, we search for the olm.package object, then the value inside of that.
This also shouldn't really be used outside of our specific use case which is stringing together the custom tasks/pipelines so far which we control the inputs to.
There was a problem hiding this comment.
The order of the keys shouldn't matter
Would only matter if value came before type as in my example. That does seem unlikely though
There was a problem hiding this comment.
Updated to use your two step recommendation! PTAL when you have a minute, thank you!!
There was a problem hiding this comment.
LGTM if the potential shortcomings work for you. Alternatively, could save the output of opm render in a shared directory and extract the version in the next step with jq (using an image that has jq installed, e.g. quay.io/konflux-ci/appstudio-utils:ab6b0b8e40e440158e7288c73aff1cf83a2cc8a9@sha256:24179f0efd06c65d16868c2d7eb82573cce8e43533de6cea14fec3b7446e0b14)
this is needed to string together the dynamic bundle generation task to
the catalog generation task. The bundle ref is stored as a param in the
catalog pipeline (nudged), we need to fetch the version that was put in
the bundle to add an entry into the catalog with it.
```
❯ opm render ${BUNDLE_IMAGE}
{
"schema": "olm.bundle",
"name": "osd-example-operator.v4.16.0-28f1940",
"package": "osd-example-operator",
"image": "quay.io/redhat-user-workloads/oeo-cicada-tenant/osd-example-operator/bundle@sha256:5485652c17ab8ac5cf44452f67e15ed06dc9d2d7d67a7f079e76248878d5370e",
"properties": [
{
"type": "olm.gvk",
"value": {
"group": "managed.openshift.io",
"kind": "Example",
"version": "v1alpha1"
}
},
{
"type": "olm.package",
"value": {
"packageName": "osd-example-operator",
"version": "4.16.0-28f1940"
}
},
...
❯ opm render ${BUNDLE_IMAGE} \
| sed -n '/"type": "olm.package"/,/}/p' \
| sed -n 's/.*"version": "\(.*\)".*/\1/p'
| tr -d '\n'
4.16.0-28f1940
```
https://github.com/openshift/osd-example-operator/blob/971083313e7eb4c7383c04a2067ac80ae535946c/.tekton/catalog-push.yaml#L176-L211
https://issues.redhat.com/browse/SDCICD-1358
Signed-off-by: Brady Pratt <[email protected]>
|
I think the |
|
/retest |
|
/retest @MartinBasti 👀 do you understand the CI failure? I'm not able to see the pipeline logs |
|
/retest |
|
Hey @MartinBasti, are you able to add this to the merge queue? |
this is needed to string together the dynamic bundle generation task to
the catalog generation task. The bundle ref is stored as a param in the
catalog pipeline (nudged), we need to fetch the version that was put in
the bundle to add an entry into the catalog with it.
https://github.com/openshift/osd-example-operator/blob/971083313e7eb4c7383c04a2067ac80ae535946c/.tekton/catalog-push.yaml#L176-L211
https://issues.redhat.com/browse/SDCICD-1358
Example pipeline
Pipeline output
Signed-off-by: Brady Pratt [email protected]