centos – OS seems to be unchanged after issuing `dnf system-upgrade reboot`

I’m hoping to upgrade from CentOS Stream 8 to CentOS Stream 9.

I’ve issued these commands

dnf install dnf-plugin-system-upgrade

dnf system-upgrade download --releasever=9 --allowerasing  --skip-broken --nobest

(I added skip-broken and nobest due to it not being able to find a suitable nginx package to install)

After that it seemed to run without issue, outputting:

Transaction saved to /var/lib/dnf/system-upgrade/system-upgrade-transaction.json.
Download complete! Use 'dnf system-upgrade reboot' to start the upgrade.
To remove cahced metadata and transaction use 'dns system-upgrade clean'
The downloaded packaged were saved in cache until the next successful transaction.
You can remove cahced packages by executing 'dnf clean packages'.

I issues the recommended command:

dnf system-upgrade reboot

I was expecting it to churn away for a while, and then reboot, but it rebooted immediately instead. Having googled it I think that’s normal.

It took a few minutes to come back online (certainly longer than a normal reboot), but now that it has the first thing I tried was

cat /etc/redhat-release

and it shows I’m still on 8, not 9.

That json file it mentioned no longer exists, but I do have system-upgrade-state.json instead, with an appropriate timestamp.

Its contents (doesn’t tell me anything useful, I think?):

{
    "destdir": null,
    "distro_sync": true,
    "download_status": null,
    "enable_disable_repos": [],
    "gpgcheck": true,
    "gpgcheck_repos": [
        "appstream",
        "baseos",
        "debuginfo",
        "extras",
        "extras-common",
        "ha",
        "media-baseos",
        "media-appstream",
        "nfv",
        "powertools",
        "rt",
        "resilientstorage",
        "baseos-source",
        "appstream-source",
        "powertools-source",
        "extras-source",
        "ha-source",
        "rt-source",
        "resilientstorage-source",
        "nfv-source",
        "epel",
        "epel-debuginfo",
        "epel-source",
        "epel-modular",
        "epel-modular-debuginfo",
        "epel-modular-source",
        "epel-next",
        "epel-next-debuginfo",
        "epel-next-source",
        "epel-next-testing",
        "epel-next-testing-debuginfo",
        "epel-next-testing-source",
        "epel-testing",
        "epel-testing-debuginfo",
        "epel-testing-source",
        "epel-testing-modular",
        "epel-testing-modular-debuginfo",
        "epel-testing-modular-source",
        "google-cloud-sdk"
    ],
    "module_platform_id": null,
    "repo_gpgcheck_repos": [],
    "state_version": null,
    "system_releasever": "8",
    "target_releasever": "9",
    "upgrade_command": null,
    "upgrade_status": null
}

What do I do now?

edit:

Hopefully this wasn’t a terrible idea. What I’ve done now is to repeat the same thing, and hope for the best…. So far it’s been unreachable for 30 minutes this time. Either it’s a good sign that it’s doing a lot behind the scenes… or it’s dead. Time will tell.