KVM Host installation¶

KVM Host Requirements¶

Hardware Requirements

Industry standard Intel/AMD Servers that supports virtualization

Ethernet Port for IPMI/iDRAC/iLO OOB (required for PAT)

Ethernet Port for Management Bus

Ethernet Port for Private Bus

Operating System Drive (preferably 2 drives in RAID 1)

VM Storage Drive (preferably multiple drives in RAID10)

Software Requirements

Ubuntu 20.04 LTS (GNU/Linux 4.15.0-65-generic x86_64)

KVM Host Installation¶

Server IPMI/iDRAC/iLO configuration:
- Assign the OOB IP address (10.<pod_number>.0.x/16, x!= 0,1,2,3,4,5,6,254,253 and make sure x is not used already)
- Set user root password
- Enable virtualization
- If LED screen kvm(x).{regionname}
Server RAID Configuration:
- Boot server into RAID configuration and create RAID volume(s):
- RAID1 - OS (at least 200GB)
- RAID10(Preferably) - VMIMAGES (The rest available storage)
Operating System Install:
- Assigning the correct management IPv6 address (<p>::30::{x}/64, x in 1-ffff ).Set DNS server to a public IPv6 DNS server address
- Set username as: administrator
- Set password for administrator to Network Password supplied by the PAT or Region POD owner.
- Set hostname to: kvm{x}-{regionname}-{organization url} (x is the number starting with 1 to ffff)
- Set Partition to: Guided - Use entire disk and set UP LVM and install on the 200GB volume
- In the software selection tab select: OpenSSH Server
- Finish installation as per prompts
Update the server:
- Login to the host as administrator, become root and patch the operating system
```
sudo su
apt update && apt upgrade -y
```

Install and run KVM server roles and features

Install packages

apt install qemu qemu-kvm libvirt-daemon-system libvirt-clients bridge-utils virtinst -y

Start the service
```
service libvirtd start
```

Configure logical volume on the VMIMAGES RAID 10 volume
- List logical device name
```
lsblk
```
- To clear a drive from the existing file systems, if any, use
```
wipefs -a /dev/sdb
```
- Create a physical device
```
pvcreate /dev/sdb
```
- Create volume group
```
vgcreate vmimages /dev/sdb
```
- Create logical volume
```
lvcreate -l 100%FREE -n lvm vmimages
```
- Format volume
```
mkfs.ext4 /dev/vmimages/lvm
```

Append /etc/fstab file and mount logical volume

echo /dev/mapper/vmimages-lvm  /var/lib/libvirt/images   ext4    defaults 0 0 >> /etc/fstab && sudo mount -a

Change the logical name of the “private” interface (usually second interface) to “cloud0”

Find interface MAC address
```
lshw -C network | grep -n2 -e serial
```

Create new file

nano /etc/udev/rules.d/70-persistent-net.rules

Add the following to the file using the MAC address from the first step

SUBSYSTEM=="net", ACTION=="add", DRIVERS=="?*", ATTR{address}=="mac:address:goes:here", ATTR{dev_id}=="0x0", ATTR{type}=="1", NAME="cloud0"

Flush the rules

ln -sf /dev/null /lib/udev/rules.d/80-net-setup-link.rules

Install NFS Common
```
apt install nfs-common -y
```

Create ISOs directory and mount NFS drive to /var/lib/libvirt/ISOs

Create a directory
```
mkdir /var/lib/libvirt/ISOs
```

Append /etc/fstab file and mount NFS share

echo robot.{regionname}.{organization url}:/etc/cloudcix/robot /var/lib/libvirt/ISOs nfs defaults,user,exec  0  0 >> /etc/fstab && mount -a

Create backup directories and mount NAS backup repositories

Create directories

mkdir /mnt/backup-p && mkdir /mnt/backup-s && mkdir /mnt/kvm_backups

Append /etc/fstab file and mount NFS shares

echo truenas{x}.{regionname}.{organization url}:/mnt/pool/kvm_backups /mnt/kvm_backups nfs defaults,user,exec  0  0 >> /etc/fstab  && \
echo truenas{x}.{regionname}.{organization url}:/mnt/pool/kvm /mnt/backup-p nfs defaults,user,exec  0  0 >> /etc/fstab  && \
echo truenas{x}.{regionname}.{organization url}:/mnt/pool/kvm-1 /mnt/backup-s nfs defaults,user,exec  0  0 >> /etc/fstab && \
mount -a

(where x is the number starting with 1 to ffff of the NAS server)

Install and configure NTP client and set to the common NTP server so that the time on the host is synced across the Region
- Install the package
```
apt install ntp -y
```
- Modify ntp.conf file, nano /etc/ntp.conf and put your preferred NTP server details under “# Specify one or more NTP servers.” section
```
server 2.ie.pool.ntp.org prefer iburst
```
- Restart the service
```
systemctl restart ntp
```
Set timezone to UTC
```
timedatectl set-timezone UTC
```
Due to an abnormal behaviour on some servers during netplan-apply (https://bugs.launchpad.net/ubuntu/+source/netplan.io/+bug/1962095) the following fixes this:
- Enable networkd.socket
```
systemctl enable systemd-networkd.socket
```
- Start networkd.socket (might fail to start if it already running)
```
systemctl start systemd-networkd.socket
```
Update SSH Sever options
- Open “sshd_config” file
```
sudo nano /etc/ssh/sshd_config
```
- Look for the “MaxStartups” option and set it equal to the number of Robot workers in the Region. The default number of Robot workers is 25.
```
MaxStartups  25
```
- Look for the “MaxSessions” option and set it equal to the number of Robot workers in the Region. The default number of Robot workers is 25.
```
MaxSessions  25
```
- Look for the “UseDNS” option and set it to “no” to disable DNS lookups when Robot tries to ssh into a host
```
UseDNS no
```
- Restart SSH service for changes to take effect
```
sudo systemctl restart ssh
```

Increase IPv6 Routing cache

sudo sysctl -w net.ipv6.route.max_size=32768

Add routing to Robot docker containers
- Open “/etc/netplan/00-installer-config.yml” in nano
- Add the following route under the management interface, telling the host to route traffic for Robot’s docker network through the Region Appliance. :
```
{management_interface}:
  routes:
    - to: "{p}:d0c6::/64"
      via: "{p}::6000:1"
```
- Save the file, exit nano, and run the following command to apply the changes:
```
sudo netplan apply
```
Place Robot’s Public Key on the host into the .ssh directory
- Connect to the Region appliance. The Region appliance password can be obtained from the PAT owner:
```
ssh administrator@{ipv6:6000:1}
```
- From appliance, copy SSH keys to the host
```
ssh-copy-id -i /home/administrator/.ssh/id_rsa.pub administrator@{kvmhostiPV6}
```
  NOTE: If system is running a “legacy” Region:
  - Change user from root to administrator: sudo su administrator
  - Create (if doesn’t exist) .ssh directory under /home/administrator directory:
    
    mkdir .ssh
  - Copy “authorized_keys” file from /var/lib/libvirt/ISOs/Robot_SSH_Key to /home/administrator/.ssh/:
    
    cp /var/lib/libvirt/ISOs/Robot_SSH_Key/authorized_keys /home/administrator/.ssh/

Copy backup script

cp /var/lib/libvirt/ISOs/KVM_Backup_Script/kvm-backup /etc/cron.weekly

Reboot the server and do an overall final check
- Become root and authenticate
```
sudo su
```
- Check if the “Private” interface is named “cloud0”
```
lshw -c network | grep -n3 cloud0
```
- Check if the following directories are mounted
  - /var/lib/libvirt/images
  - /var/lib/libvirt/ISOs
  - /mnt/backup-p
  - /mnt/backup-s
  - /mnt/kvm_backups
Add data about the server into IaaS via the Franchisee App following the steps here. This is so Robot can find out where it can build infrastructure.

Table Of Contents

Contents

Previous topic

Next topic

KVM Host installation¶

KVM Host Requirements¶

KVM Host Installation¶