manifests: Serve oc from within the cluster #177
Conversation
We need to support the ability to download clients (oc) from the cluster when an install is successful. With this commit, I'm adding a new Deployment for serving the binary from the cli image referenced from the release image. That way, the provided oc is the appropriate version for the current cluster version. I've set 'replicas' to two to support continued availability during rolling updates. The Python here-document approach is based on openshift/ci-operator@95a41a39 (Use a multithreaded python server to serve RPMs, 2018-06-25, openshift/ci-operator#35). The script creates a scratch directory to serve from, and populates it with the oc binary (via a symlink). It also creates zip and tar versions, so users can do things like: $ wget https://downloads-openshift-console.apps.wking.devcluster.openshift.com/amd64/linux/oc.zip $ unzip oc.zip instead of: $ wget https://downloads-openshift-console.apps.wking.devcluster.openshift.com/amd64/linux/oc $ chmod +x oc if they prefer. We don't have to clean up the scratch directory in the script, because it will be reaped as part of container cleanup. Moving forward, we'll want to replace the cli image with a cli-downloads image that includes builds for other platforms and architectures, but I'm punting on that for now. [1]: openshift/ci-operator@95a41a3
openshift-ci-robot
added
the
size/L
|
@spadgett fyi |
|
/lgtm heredoc will always bother me but I can trust its the right call. 😄 |
openshift-ci-robot
added
lgtm
|
Contemplating if I should follow and split |
Does the CVO walk subdirs?
Maybe if we can't find a way to serve all downloads from a single container? |
|
@wking I'm looking for info about subdirs now. I see at least one operator with subdirs, but I dont know if CVO manages it. If not, perhaps I'll do a I'm guessing we may also get |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: benjaminapetersen, wking The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
I'll run my reapers. |
|
/retest Please review the full test history for this PR and help us cut down flakes. |
|
🎉 |
5 participants
We need to support the ability to download clients (
oc) from the cluster when an install is successful. With this commit, I'm adding a new Deployment for serving the binary from the cli image referencedfrom the release image. That way, the provided
ocis the appropriate version for the current cluster version. I've setreplicasto two to support continued availability during rolling updates.The Python here-document approach is based on openshift/ci-operator@95a41a39 (openshift/ci-operator#35). The script creates a scratch directory to serve from, and populates it with the
ocbinary (via a symlink). It also creates zip and tar versions, so users can do things like:instead of:
if they prefer. We don't have to clean up the scratch directory in the script, because it will be reaped as part of container cleanup.
Moving forward, we'll want to replace the cli image with a cli-downloads image or similar that includes builds for other platforms and architectures, but I'm punting on that for now.