Static IP addresses with kickstart installation of CentOS 7

Question

The following error is being thrown when a kickstart file is used to install a CentOS 7 guest virtual machine with static IP networking:

[3.835698] dracut-cmdline[81]: parse-kickstart ERROR: 
    'network --bootproto=static --ip=12.34.567.8aa --netmask=255.255.255.248 --gateway=12.34.567.8bb --nameserver=xx.xx.xx.xx,xx.xx.yy.yy': 
    missing --device

I suspect that the problem is that bridge networking with static IP has not been set up at the host to replace the default NAT configuration. But what specific commands need to be typed in order to set resolve this error?

---

Kickstart file:

---

The kickstart file is:

#version=RHEL7
# System authorization information
auth --enableshadow --passalgo=sha512

# Run the Setup Agent on first boot
firstboot --enable
ignoredisk --only-use=sda
# Keyboard layouts
keyboard --vckeymap=us --xlayouts='us'
# System language
lang en_US.UTF-8

# Network information
network  --device=eno1 --onboot=on --activate
network  --bootproto=static --ip=12.34.567.8aa --netmask=255.255.255.248 --gateway=12.34.567.8bb --nameserver=xx.xx.xx.xx,xx.xx.yy.yy
network  --hostname=localhost.localdomain
# Root password
rootpw --iscrypted someLongHashedPassword
# System timezone
timezone someTimeZone --isUtc --nontp
user --name=someUserName --password=someLongHashedPassword --iscrypted --gecos="someUserName"
# System bootloader configuration
bootloader --append=" crashkernel=auto" --location=mbr --boot-drive=sda
# Partition clearing information.  Erases all partitions from the sda drive.
clearpart --all --initlabel --drives=sda
# Disk partitioning information
part pv.204 --fstype="lvmpv" --ondisk=sda --size=1902212
part /boot/efi --fstype="efi" --ondisk=sda --size=200 --fsoptions="umask=0077,shortname=winnt"
part /boot --fstype="xfs" --ondisk=sda --size=500
volgroup centos --pesize=4096 pv.204
logvol /  --fstype="xfs" --grow --maxsize=51200 --size=1024 --name=root --vgname=centos
logvol /home  --fstype="xfs" --size=230400 --name=home --vgname=centos
logvol swap  --fstype="swap" --size=7808 --name=swap --vgname=centos

%packages
@base
@compat-libraries
@core
@debugging
@development
@network-file-system-client
@remote-system-management
@security-tools
@smart-card
@virtualization-hypervisor
@virtualization-platform
@virtualization-tools
@virtualization-client
kexec-tools

%end

%addon com_redhat_kdump --enable --reserve-mb='auto'

%end

---

virt-install Command:

---

For reference, the virt-install command that triggers the install is:

virt-install \
   --name=public-centos7 \
   --disk path=/dev/mapper/centos-fifth,size=241 \
   --graphics none --vcpus=1 --memory=2048 \
   --location /tmp/CentOS-7-x86_64-Minimal-1611.iso \
   --network bridge=virbr0 --os-type=linux --os-variant=rhel7.0 \
   --initrd-inject=/tmp/vm.ks \
   --extra-args "ks=file:/tmp/vm.ks console=ttyS0"

---

Current Config:

---

Also, brctl show on the host machine gives:

[root@remote-host ~]# brctl show
bridge name     bridge id               STP enabled     interfaces
virbr0          8000.525400c4a345       yes             virbr0-nic
                                                        vnet0

---

Adding --device=eno1

---

Per @thrig's suggestion, I changed the offending line of the kickstart file to become:

# Network information
network  --onboot=on --activate
network  --bootproto=static --ip=12.34.567.8aa --netmask=255.255.255.248 --gateway=12.34.567.8bb --nameserver=xx.xx.xx.xx,xx.xx.yy.yy --device=eno1
network  --hostname=localhost.localdomain  

This seems to have resolved the error. But I am not yet certain because I am still resolving downstream problems.

Solution

The error message indicates "missing --device" so a good thing to try is associating a device with the network configuration line:

network  --bootproto=static --ip=... --device=eno1

This may be problematical if the device names are coming up with some unknown name depending on their PCI position, though there are other options to control that (e.g. PXE booting things with ksdevice=eth0 net.ifnames=0 biosdevname=0 kernel arguments). In particular, the redhat "Installation Guide" docs indicate that the device name really should be specified:

Note that this is considered deprecated behavior; in most cases, you should always specify a --device= for every network command. The behavior of any subsequent network command in the same Kickstart file is unspecified if its --device= option is missing.

Comments

[RabT]

Thank you and +1 for quickly enabling me to move on to downstream issues. I have re-framed the downstream question after more careful examination and research/tinkering. Are you willing to comment? Here is the link: unix.stackexchange.com/questions/352181/…

[Centimane]

Also worth noting from the documentation you can use link as the device so as to avoid hard coding the interface name, and it will select the first NIC with an active link.