Please check redis and api-server:
systemctl status api-server redis
Maybe try again without reboot after nmtui?
1 Like
Checcking the status of api-server and redis, they are both running.
You want me to start from clean image and do nmtui and then just restart network manager (not VM) and see what happens?
It’s just a guess…here the qcow2 image worked on Proxmox without issues but I didn’t reboot.
The root user seems to be a disallowed user by default in Cockpit.
Just edit /etc/cockpit/disallowed-users and comment out root to make the login work.
1 Like
But admin ALSO doesn’t login there.
I will start from a new image.
It is weird that I am the only one reporting these things (?)… I am only following the install docs.
2 Likes
admin is just a NS8 user, not a Cockpit user AFAIK.
2 Likes
So is, by default, ANY user allowed in Cockpit when NS8 first starts?
(my main issue of course remains that I never saw any “first start” script running nor have access to cluser admin - I’ll get back on that when I begin with a fresh VM image)
2 Likes
Any system user is allowed to access cockpit but only with restricted access if not in sudoers file.
See also Privileges and Permissions
2 Likes
Back to the main point.
- http(s)://[ip I defined in nmtui]/cluster-admin does nothing (fails).
I have (exactly) the same experience as OP.
What to do next? Is there any solution or explanation or advice what to try next?
Thank you.
Unless something’s changed, you need a trailing slash: https://server_ip/cluster-admin/. Have you tried it that way?
Why would that make a difference, Dan, I am really interested.
It’s been years since I last heard it mattered.
All web servers I know are smart enough to understand if the URL ends in a file or a folder.
I can’t really speak to the “why,” but that was definitely required when I first tested NS8.
Well I can verify it doesn’t work with or without - in my case.
But if that makes a difference, it means that NS8 web server is not properly configured I would say.
1 Like
So I started from new image.
I edited network settings and enabled them without rebooting.
cluster-admin (with or without trailing slash) works this time.
I rebooted, it still works.
Only difference from first time is that in the first time the image didn’t boot 3-4 times because it was not set as qcow2 (shouldn’t be related to the issue) and that last time I edited the network settings and then rebooted (which DID make the network settings change). I never saw cluster-admin.
It seems weird that this little difference could affect the whole server.
Also I would DEFINITELY expect pre-built images to force me edit the IP or inform me of current IP at least (as I could have arranged things from DHCP side) when they first boot. It is not clear in the docs (yes it is “more or less mentioned” with “if you have network issues” etc. - but definitely needs to be cleaner) what you need to do.
Thanks for all the help.
PS. Rocky Linux cockpit, still doesn’t allow login of root by default. I find it a bit weird, but I guess that is Rocky Linux issue or policy.
4 Likes
Shoutout to whoever handles the documentation.
There is still no mention in fresh install using pre-made images, to configure the network first!
Network of the pre built image is configured with DHCP: it is cloud-init fallback method. If it fails and the host is unreachable, follow the hint from
https://docs.nethserver.org/projects/ns8/en/main/install.html#post-installation-steps
Thanks. Good enough I guess.
That said,
a) Is it good practice for a server to use DHCP?
b) Should probably be even more clear. I think in the first console login, above login prompt, it does NOT hint on how to get web access (like it does in subsequent - after initial configuration - boots).
In other words it doesn’t show current IP when user needs it the most: The first time!
Maybe this needs to change?
Yes, that’s how it works. After several hours of pointless attempts. Nevertheless, there should be a simpler solution that you can use without having to set up the system from scratch once the child has fallen into the well.
1 Like
According to the team (?) the pre-made images suck and are “quick n dirty” and should prefer install over existing distro.
I hope this changes. Really having a ready-made ISO is vital for SB deployments.
(plus it is a more controlled environment for debugging)