- 3,714,551 Users
- 2,242,575 Discussions
- 7,844,931 Comments
Forum Stats
Discussions
Categories
- Industry Applications
- 3.2K Intelligent Advisor
- Insurance
- 1K On-Premises Infrastructure
- 356 Analytics Software
- 32 Application Development Software
- 1.7K Cloud Platform
- 700.5K Database Software
- 17.4K Enterprise Manager
- 7 Hardware
- 164 Infrastructure Software
- 87 Integration
- 51 Security Software
Error: Connection actifaction failed: No suitable device found for this connection (device virbr0 no

=================================================================================
The original question was :
How to prevent a specific kernel from being removed on Oracle Linux 7.9 ?
I just updated my current kernel (5.4.17-2011.6.2.el7uek.x86_64) to a newer version (5.4.17-2036.103.3.1.el7uek.x86_64) and the new kernel did not detect my network card.
I rebooted with the old one and the network card is detected again.
Question : How can I prevent Oracle linux from removing a specific kernel ( in this example 5.4.17-2011.6.2.el7uek.x86_64 ) ? I already raised installonly_limit = 5 and have a backup of / so I can always restore it. I will verify the number of kernel at each update so it won't go beyond 5 . I just like to have an extra precaution so I don't have to restore in case I forget to verify the number of kernels.
==================================================================================
Once I got the answer to my kernel issue I decided to change the question title so it would be more usefull to users ( than the original title )
Best Answer
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
Could you please run:
# dracut --add-drivers realtek -f /boot/initramfs-5.4.17-2036.103.3.1.el7uek.x86_64.img 5.4.17-2036.103.3.1.el7uek.x86_64
And then reboot into that kernel? If that works, add:
drivers+="realtek"
To the
/etc/dracut.conf
so that it is always included in future.
Answers
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
I'd rather ensure that your network card keepss working when you upgrade. What kind of network card is it?
-
This is an integrated network from my motherboard
lspci | grep -i "ethernet"
06:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller (rev 02)
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
It surprises me that it disappears after you upgrade. Does it disappear from
lspci
completely? If so, could you provide the output fromdmesg
andlspci
from the working kernel and the non-working kernel?You don't have enough posts to attach a file, so I suggest using a GitHub Gist or Pastebin.
-
I just verified and I was wrong. It did not disappear, it's still there.
The actual error that I got when I attempted to start the network interface is
ifup enps6s0 # this is my network interface
Error: Connection actifaction failed: No suitable device found for this connection (device virbr0 not available because profile inot compatible with device (mistmaching interface name))
( When I saw the error I presume wrongly that the device disappear ).
Do you still want to see dmesg and lspci ?
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
No, but I would like to see the output of
nmcli con show
for each kernel. What's happening is the newer kernel is naming the device differently and NetworkManager isn't starting it. I'd like to work out why its name is different. -
-- Kernel that works (5.4.17-2011.6.2.el7uek.x86_64)
NAME UUID TYPE DEVICE
enp6s0 46acbda0-422e-48da-9761-4e9ba0706a93 ethernet enp6s0
virbr0 900f0500-4aa1-4aa9-9a5d-3c52411904ca bridge virbr0
-- Kernel with issue (5.4.17-2036.103.3.1.el7uek.x86_64)
NAME UUID TYPE DEVICE
virbr0 48d00365-551a-4c87-b605-897277545da2 bridge virbr0
enp6s0 46acbda0-422e-48da-9761-4e9ba0706a93 ethernet --
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
Ok, the NIC still has the same name, so I will need to see the
dmesg
output after all. Hopefully that will tell us why the NIC isn't starting. -
I just uploaded ol81.7z which contains dmesg_ok.out (kernel with ok network) and dmesg_bug.out
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
The NIC isn't in the
dmesg
output at all on the non-working server, which makes me suspect that when theinitramfs
was built during the kernel install, something weird happened. Can you check that you have enough free space in/boot
to create newinitramfs
files?If you do have limited disk space, I suggest removing a couple of the older kernels so that the installation is able to build the
initramfs
successfully.Alternatively, if you have plenty of space, add
rdinitdebug
to the kernel boot line so we can get some debug output from the that phase of the boot. -
I have plenty of space on /boot ( / and /boot are on the same partition with 98G free ).
Once I added rdinitdebug to kernel boot line and reboot what do you want me to upload ?
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
I've actually pinged a couple of folks internally who know this stuff bettr than I do to see if they can help out.
-
Avi Miller-Oracle Senior Solution Architect, Oracle Cloud Infrastructure Developer Adoption Melbourne, AustraliaPosts: 4,727 Employee
Could you please run:
# dracut --add-drivers realtek -f /boot/initramfs-5.4.17-2036.103.3.1.el7uek.x86_64.img 5.4.17-2036.103.3.1.el7uek.x86_64
And then reboot into that kernel? If that works, add:
drivers+="realtek"
To the
/etc/dracut.conf
so that it is always included in future. -
It works. Thank you very much.