WIP bootstrap: change bootstrap host to replace itself with machine-os-content too #2559
Conversation
openshift-ci-robot
added
the
size/S
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: vrutkovs The full list of commands accepted by this bot can be found here.
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
|
||
| mkdir bin/ | ||
| bootkube_podman_run \ | ||
| --volume "$PWD/bin:/host/usr/local/bin:z" \ |
There was a problem hiding this comment.
This will relabel the host's /usr/local/bin - dangerous. Let's instead do something like:
hostmcd=/usr/local/bin/machine-config-daemon
bootkube_podman_run --entrypoint=sh "${MACHINE_CONFIG_OPERATOR_IMAGE} cat /usr/bin/machine-config-daemon >${hostmcd}
chmod a+x ${hostmcd}
restorecon ${hostmcd}
|
Commit message title is overlong, and the body could use more information and links; how about:
|
vrutkovs
force-pushed
the
pivot-bootstrap-node
branch
from
f8ea35d
to
8fb4114
Compare
Currently every machine instance we launch uses the same "bootimage", including the bootstrap host. However, everything except bootstrap (i.e. control plane and workers) replace their OS content with the machine-os-content from the release payload before joining the cluster. For more information, see: https://github.com/openshift/machine-config-operator/blob/master/docs/OSUpgrades.md This changes the bootstrap host to do the same, which will help avoid issues from "bootimage drift". Closes: openshift#2542
vrutkovs
force-pushed
the
pivot-bootstrap-node
branch
from
8fb4114
to
a661f2e
Compare
|
/hold this needs an enhancement before we think about merging it. |
openshift-ci-robot
added
the
do-not-merge/hold
I don't think this enhances anything really, its fixes the situation where bootstrap node runs a different version of kubelet / crio until bootimages are bumped in the installer. In openshift/enhancements#78 (comment) Clayton considers this to be a bug, and so do I. I don't mind holding this for now, however later on RHCOS would stop producing boot images for .z releases. That might cause various bootstrap issues during new cluster installs |
|
I agree with abhinavdahiya that this is a notable architectural change that requires an enhancement; it's something that a lot of OpenShift developers and some advanced users will end up needing to understand. I also think we already have an enhancement for this in openshift/enhancements#78 right? |
78 is a bit different, pivoting bootstrap in context of OKD-on-FCOS is the architectural decision to avoid maintaining a different FCOS stream. That is not the situation we have in RHCOS, so I don't think enhancement#78 covers this case. |
vrutkovs
changed the title
vrutkovs
changed the title
openshift-ci-robot
added
the
do-not-merge/work-in-progress
|
/retest |
|
@vrutkovs: The following tests failed, say
Full PR test history. Your PR dashboard. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
|
Closing this for the time being, when we have an enhancement for this change we can re-open linking to it. /close |
|
@abhinavdahiya: Closed this PR. In response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
…pivot before proceeding
This ensures bootstrap node uses latest crio/kubelet from oscontainer. Bootimages may not be bumped frequent enough to reflect oscontainer updates
TODO:
Fixes #2542