Attempted to install NethServer 8. Tried with Debian 11 and CentsOS 9 Stream. Both times failed to connect to NethServer 8 web interface. Tried the third time with CentsOS 9 Stream and installed everything, including GUI. This time I was able to connect to NethServer 8 web interface, create a master node and add a module.
Previous versions of NethServer did all the work in providing a working system. Now we are expected to install the base server OS without any guidance. This is no longer supporting the KISS principle.
My question, what are the minimum requirements in terms of services and applications to install the base server OS? DNS, firewall, mail server and etc. Looking for guidance to setup the base server OS.
IMVHOâŚ
Working/installed OS. Static LAN IPv4 address, at least one DNS server configured (as part of TCP/IP Stack)
For the working hardware, take your time to check list for CentOS Stream and Debian 11.
You realize that youâre talking about Alpha-level software, right? This is pre-pre-release software, and itâs to be expected that documentation is going to be absent. Your questions are perfectly valid, but your complaint above, IMO, is very premature.
To the substance of your question, Iâve been able to run it with a base Debian 11 install. The only additional software Iâve installed is the guest agent for my hypervisor, and stuff dealing with SSH certificates, both of which are part of the Debian template Iâve set up.
Installing Debian 11 or CentsOS 9 Stream Server offered many choices. The first 2 attempts were installed with no options selected, installed Nethserver 8 successfully. But unable access to the web site to with clusters. The last attempt was installed using CentsOS 9 Stream Server with all options selected. The suggestion was to install Nethserver 8 with no GUI. All I am asking is what are the minimum server options to be installed? Or does it work with no options selected? Is there a firewall in the way?
Iâd agree that should be part of the packages to install, but really youâd want to have installed it as part of the base OS. But yes, firewall should allow access on green.
With CentOS 9, I set the âsoftware selectionâ to âminimum installâ and added âguest agentsâ (because Iâm running it as a VM). I didnât select anything else to install. I set a root password, and checked the box to allow root login with password via SSH. I set a hostname, and then ran the install.
Once I rebooted, I ran (per the docs) curl https://raw.githubusercontent.com/NethServer/ns8-core/main/core/install.sh | bash. Once that finished, I browsed (using Firefox; Chrome gave SSL errors) to https://192.168.1.227/cluster-admin/ and, after accepting the self-signed cert, got this page:
Logged in using the credentials given in the terminal (admin/Nethesis,1234), tell it to create a cluster, change the admin password, and get this page:
There are many ISO, variants and VM images to start from, for both distros. I usually run Debian 11 Digital Ocean official image, and CentOS Stream 9 âgeneric cloudâ image.
So itâs important to share any experience with other installation options!
Installing ssh, and other basic requirements setup can be added to the install.sh script: Pull Requests on GitHub are welcome! Please remember to explain why the Pull Request is opened
I replied above that the complaint about failing the KISS principle is premature, but as I think further about it, consider this: youâre installing, and presumably expecting to administer, a server thatâs designed to be exposed to the Internet. I donât think expecting the admin to be able to step through a Debian or CentOS installer (both of which are really pretty simple to run through) is in any way excessive, particularly when thereâs nothing beyond the minimum install thatâs required.
As Iâve shown, it runs fine (did it this afternoon) on a minimum CentOS 9 installation (installed from the ISO downloaded from centos.org).
So Debian 11 or CentOS 9 are the (currently) preferred distros.
But is one of the two⌠âa bit more preferredâ?
I mean for someone that really has not fav or personal preference, which of the two will more easily result in a working NS8 alpha environment? Preferably virtualized.
IMVHO choice should be made by the sysadmin favourite flavour.
Bullseye should have more or less 3 years to update (release august 2021) before EOL.
Currently debian supports more other versionsâŚ
One ELTS version (Jessie, EOL July 2022) and one LTS (Stretch, same EOL). Buster also is going to be EOL at September 2022.
CentOS should be a rolling release, so currently no EOL/version change is expected, at least as current project claims.
At NS8 current project state, my personal pick would be a rolling release as testbench. It will ease the progress of the installation during the steps and time, not requesting me to change release âthatâ soon.
Most of my future choices about the underlying Linux distro could dramatically change by the steps and costrains of backup/restore/duplicate/move procedures for the containters; Iâd be rather confused and puzzled if a âcross flavaâ container move/migration/restore would not be possible, due to the project choices.
Come from Google. Who thinks that Apple is some kind of amateur at the art of âwe are the internet, and the world should faster to acknowledge itâ.
