You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is a feature request. When facing network issues, it is useful to log in to cluster nodes via IPMI to debug. This should ideally be done on two levels:
Prior to installation by modifying the installation media to debug installation failures.
Post-installation by creating MachineConfig manifests.
The latter could be done by users themselves by supplying a MachineConfig manifest. The following will work on SNO, for other clusters another label targetting all cluster nodes would have to be used.
An aside: applying this manifest on recent OCP versions will not even reboot the cluster node.
I believe this could be done in one of two ways:
By supplying ignition files to be embedded into the installation media and adding MachineConfig manifests from a manifests directory the way it is done by the bootstrap-in-place-poc.
By letting users to configure a user/password combination.
The former approach is more flexible and will solve many other issues/requests users of this tooling have, the latter is probably more novice user friendly.
Perhaps we could start by copying the approach taken by bootstrap-in-place-poc and then add the user/password option as an abstraction on top of it.
Thoughts?
The text was updated successfully, but these errors were encountered:
This is a feature request. When facing network issues, it is useful to log in to cluster nodes via IPMI to debug. This should ideally be done on two levels:
I believe the first could be achieved by using the
quay.io/coreos/coreos-installer:release
. Here is an example how to do this.The latter could be done by users themselves by supplying a MachineConfig manifest. The following will work on SNO, for other clusters another label targetting all cluster nodes would have to be used.
An aside: applying this manifest on recent OCP versions will not even reboot the cluster node.
I believe this could be done in one of two ways:
The former approach is more flexible and will solve many other issues/requests users of this tooling have, the latter is probably more novice user friendly.
Perhaps we could start by copying the approach taken by bootstrap-in-place-poc and then add the user/password option as an abstraction on top of it.
Thoughts?
The text was updated successfully, but these errors were encountered: