Monitoring the completion of the bootstrap phase (and installation one) of a SNO cluster times out, indicating thus a failure of the procedure, despite the cluster being deployed correctly.

The analysis so far identified the main cause in the absence of the bootstrap ConfigMap in the deployed cluster, which is the resource monitored by the openshift-install agent wait-for bootstrap-complete command to detect the completion of the bootstrap phase - note that currently also the openshift-install agent wait-for install-complete command fails as well, since internally it invokes the bootstrap-complete one.

The bootstrap ConfigMap is generated by the progress service [1], which constantly monitors the status of the bootkube service [2] (by looking for the existence of the /opt/openshift/.bootkube.done file).

The root issue is that, when the progress service decides to generate the bootstrap ConfigMap, there's no control plane available to receive the command and the operation fails silently withouth leaving any trace - since the node is almost immediatly rebooted by the assisted-installer (since it monitors independently the bootstrap completion [3]).

[1] https://github.com/openshift/installer/blob/master/data/data/bootstrap/files/usr/local/bin/report-progress.sh
[2] https://github.com/openshift/installer/blob/master/data/data/bootstrap/files/usr/local/bin/bootkube.sh.template
[3] https://github.com/openshift/assisted-installer/blob/master/src/installer/installer.go#L580