If I install the system without GUI on one computer and then install this hard drive with system on another computer, then the NM starts renaming the network cards and as a result the system only goes to the command line with dracut.And it is not possible to change the settings of network cards
You can try nmtui, it will help you config NM in GUI type.
Thanks, what can I do if the system only goes to dracut in command line?What commands does dracut have? In Сentos 6 and 7, I could simply configure the network cards manually on the command line. And here NetworkManager intervenes and prevents this from happening
sorry, I don’t know how the change of network cards could case into dracut.
make sure boot normally firstly?
This photo is the result after NM started renaming the network cards. And if you return the hard drive to the computer where the system was installed, everything boots normally.That is, NM destroys the system boot on another computer
One can “simply configure the network cards manually on the command line” on Rocky 9 too – with nmcli or nmtui.
False conclusion. NM is not the culprit there.
This is the issue. Are the hardware identical?
the issus in the photo, do not related to NetworkManager. its the disk could not be mounted.
There are a number of reasons for the unable to load hard disk error, and it has to be dealt with on a case-by-case basis for your specific situation.
You can check disk by lsblk or ls /dev/mapper inside dracut
And if I return the hard drive to the computer where the system was installed, everything boots normally. 
It’s completely clear to me what’s wrong with NM. Read in the photo “…a bug report” In general, I realized that the system is not fully functional. Maybe it will work over time. I remember Centos 6.0 also had bugs that made it impossible to stable work, but then with new releases starting from 6.2 everything started working fine.
By the way, I looked at the VzLinux website, which is also based on Rhel packages. There, el9 is a beta version. … There, version 8 is stable…Now I’m starting to understand why openpbs doesn’t create packages on el9
That site tells that their “VzLinux 9” distro is at beta stage. That does not tell anything about the quality of the RHEL 9, just that VzLinux has not put more effort on their own distro.
Their “VzLinux 8” is also released Fall 2021 and based on RHEL 8.5 content. Thus, seriously outdated.
The only “message” that I get from VzLinux website is that I should not use VzLinux distro.
When system boots the kernel does a lot and many messages are printed out. It has always been so, but the systemd starts services more in parallel than earlier inits.Therefore, the boot does not halt the moment the first error occurs. No, other services can still continue starting, until things that depend on the component that did fail try to start.
It has always been important to look at the first error messages, not the last messages that are on the screen.
Your screenshot does show that local filesystems were not found. NetworkManager is not any part of that process (unless there are iscsi-volumes). We could not see the first errors in the screenshot.
I did ask whether the two computers have identical hardware. You have not answered.
It is not “completely clear” to us what is the issue in the computer, but it is clear that NetworkManager is not part of it.
RHEL 9.0 may have had some bugs, but RHEL 9 is already 9.3. At least as good as CentOS 6.3 was. “Working fine.”
In Centos 6 and 7 there was no this stupid nmcli so the network cards could be configured manually. Debian, thank God, does not have this monster, so I finally decided to use it or ubuntu and the server with NAT is on freebsd and rinetd has already been fixed there so I can work calmly and stably what I wish for all of you too (The computers are different and one has two network cards and the other has three. But I install hard disks with Debian ,Freebsd or EndeavourOS back and forth and there are no problems , the system boots and then I just change the network card data on freebsd without GUI,-debian ,endeavour and ubuntu with GUI connect automatically
)
The nmcli is neither stupid nor smart. It is just a new tool to “change the network card data on the command line” with.
Red Hat is providing and promoting a new approach, a configuration management system named Ansible.
There are “RHEL System Roles” for Ansible. One of them can configure network for you regardless of whether you have NetworkManager or initscripts on RHEL or Debian. No need to call nmcli or whatnot explicitly.
I remember there were similar words when the Unity graphical environment was released in Ubuntu. But then after numerous complaints they abandoned it.What kind of seller says bad things about his product? 
Give users the opportunity to disconnect nmcli, and there will be no questions
That is a point, but we (you and me) are not sellers. We are users.
Red Hat is a seller. Rocky is not really a seller – many users seem to desire Enterprise Linux while not able to use RHEL. You yourself was one of them, when CentOS 5 and 6 did exists.
In other words: if you want the RHEL change, then tell that to Red Hat.
NetworkManager has been the default since RHEL 7.0. It was a tech preview in RHEL 6.
I had some legacy schemas and NetworkManager has supported them since RHEL 7.3.
It is possible that some really esoteric setups still lack support in NM, but for vast
majority of use cases it is “rock solid”. I would not use NM in compute clusters if it would
not do my bidding. It does, so I do use it. Obviously, I had to learn it first.
Red Hat has documentation on the use of NM: https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/9/html-single/configuring_and_managing_networking/index
They offer users opportunity to learn nmcli (and other UI).
You wrote:
You tell what you did:
If I install the system without GUI on one computer and then install this hard drive with system on another computer
and the result:
the system only goes to the command line with dracut
Your screenshot hints that the reason for ending up in emergency mode is because kernel
was not able to find and mount system volumes. That error must have occurred way before the
NetworkManager is starting.
Rather than showing us more actual data, you do assume that error is from NM. IMHO, your assumption that NM causes that error is false. You can test my claim:
Disable the NetworkManager.service on the “normal” system. Reboot to show that system does still start, even without NM.
Then move the drive to the other machine. Does it boot “normally”, or fail like before?
If it still fails, then NM cannot be the issue.
1 Like