[vpc][rhel-8.6][cloud-init?] Instance moving to failed state some time after provisioning

[fernflower]

I have noticed that rhel-8 s390x base cloud images with cloud-init 21.1 on board (rhel-8.5, rhel-8.6) will never reach running state but rather move to failed ~10 minutes after instance-create.

rhel-8.7 (and later releases) are fine with cloud-init 22.1 on board.

Instance details for failed rhel-8.6 don't reveal much: cannot_start_compute Can't start instance because provisioning failed. .

Upon connecting to the failed instance there is nothing alarming in journalctl or cloud-init logs.

It still feels that IBM Cloud needs something particular to finish the starting->running transition. Could you please provide any further advice how to debug / fix the matter? We'd really love to run rhel-8.6 on IBM Cloud.
Thank you!