Upgrade to 4.10.0-0.okd-2022-07-09-073606 - initial cluster version cleanup container rm issues

[stanbog]

Describe the bug

After trying to upgrade the cluster from version 4.10.0-0.okd-2022-06-10-131327 to 4.10.0-0.okd-2022-07-09-073606 , the job version-4.10.0-0.okd-2022-07-09-073606-xxxxx under namespace openshift-cluster-version failed with the following error:
rm: invalid option -- 's' Try 'rm ./-sts-xjA5wJZtz1bRCXw5w' to remove the file '-sts-xjA5wJZtz1bRCXw5w'. Try 'rm --help' for more information.

The error was generated by the cleanup container with the command:
sh -c rm -fR *
Container image: registry.ci.openshift.org/origin/release@sha256:5a748561c829c23c7413b0fe0ec8f97b4eabf5cfd1934b9e93a8a5df3f59738d

Apparently one of the folders on one of the master nodes that was trying to be deleted by the cleanup container was randomly named like:
/sysroot/ostree/deploy/fedora-coreos/deploy/6c979558a5d3679cee96bc5d5ead6c6751ff6bd8d2cc8b261d5b6393a31eb054.0/etc/cvo/updatepayloads/-sts-xjA5wJZtz1bRCXw5w

The rm command saw the - at the beginning of the folder name as a parameter and failed.
This can probably be easy fixable by re-building the cleanup container image with the command like:
sh -c rm -fR -- *

We managed to fix this by deleting the folder manually from the node.
Afterwards, the job completed and the upgrade continued.

Version

4.10.0-0.okd-2022-06-10-131327

How reproducible

100%

[vrutkovs]

That looks like a problem with CVO. Could you report it to https://issues.redhat.com/browse/OCPBUGS, component "Cluster Version Operator"