drop supermin for upload-oscontainer, use digestfile (not skopeo)#1502
Conversation
|
Tested in a dev cosa shell in PSI, and this works quickly. I also ensured we clean up our temporary storage after we're done. |
|
Tangentially: Also I investigated building the oscontainer using a proper |
In general we should avoid doing any nontrivial networking in supermin because usermode networking is slow. Further, as of OpenShift 4 it's documented how to use `buildah` inside OpenShift, so let's do that rather than supermin. Motivated by debugging why the oscontainer push in the RHCOS pipeline is 2 hours. Closes: coreos#1500
cgwalters
force-pushed
the
oscontainer-buildah-vfs
branch
from
d031f16 to
a2c4f40
Compare
|
This is only used by RHCOS right? So let’s get this in and make our pipeline sane again. Leaving /approve |
A while ago, we were using `skopeo inspect` as part of `pivot` for OpenShift 4 OS upgrades, and it turns out that enumerates all tags, which is expensive and rate limited by quay.io: openshift/pivot#51 The RHCOS oscontainer builds have been failing in `skopeo inspect` and I suspect it's related to this. It looks like nowadays podman has `--digestfile` which is exactly what we want here, so let's use it and drop the `skopeo`. This simplifies the code too.
cgwalters
changed the title
|
Folded in #1503 for simplicity of merging - while they're logically independent let's get both in at once rather than test intermediate states. |
ashcrow
requested review from
ashcrow and
dustymabe
and removed request for
miabbott and
mike-nguyen
ashcrow
left a comment
ashcrow
left a comment
There was a problem hiding this comment.
The code changes look good but one question, should cmd-upload-oscontainer be updated since we're dropping cosa oscontainer here?
You linked to code in git master, which this PR is changing right? |
🤦 yes. |
|
/lgtm |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: ashcrow, cgwalters, miabbott 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 |
5 participants
In general we should avoid doing any nontrivial networking in
supermin because usermode networking is slow. Further, as of
OpenShift 4 it's documented how to use
buildahinside OpenShift,so let's do that rather than supermin.
Motivated by debugging why the oscontainer push in the RHCOS
pipeline is 2 hours.
Closes: #1500