Add libvirt notes

config/automation/puppet.md

@@ -6,7 +6,7 @@ breadcrumbs:
 ---
 {% include header.md %}
 
-Using **Debian 10 (Buster)**.
+Using **Debian**.
 
 ## General
 

config/virt-cont/docker.md

@@ -6,7 +6,7 @@ breadcrumbs:
 ---
 {% include header.md %}
 
-Using **Debian**, unless otherwise stated.
+Using **Debian**.
 
 ## Setup
 

config/virt-cont/libvirt-kvm.md

@@ -0,0 +1,107 @@
+---
+title: libvirt & KVM
+breadcrumbs:
+- title: Configuration
+- title: Virtualization & Containerization
+---
+{% include header.md %}
+
+libvirt is a tool for managing platform virtualization like KVM and QEMU (among others).
+I'll only focus on using it with KVM (and QEMU) here.
+
+Using **Debian**.
+
+## General
+
+- Requires Intel VT or AMD-V to be enabled to function optimally.
+- Note that running commands as non-root or not in the libvirt group will generally work but will not show all VMs. etc.
+
+## Installation
+
+1. Install without extra stuff (like GUIs): `apt-get install --no-install-recommends iptables bridge-utils qemu-system qemu-utils libvirt-clients libvirt-daemon-system virtinst libosinfo-bin`
+1. (Optional) Install `dnsmasq-base` for accessing guests using their hostnames.
+1. (Optional) Add users to the `libvirt` group to allow them to manage libvirt without sudo.
+
+## Usage
+
+### Manage VMs
+
+- Show VMs: `virsh list --all`
+- Start/shutdown/reboot/kill VM: `virsh {start | shutdown | reboot | kill} <vm>`
+- Suspend/resume VM: `virsh {suspend | resume} <vm>`
+- Enter/exit console for VM:
+    - Enter: `virsh console <vm>`
+    - Exit: `Ctrl+¨` (Norwegian) or `Ctrl+]` (US)
+- Create VM:
+    - Example: `virt-install --name=example-vm --network=network=default,model=virtio --os-variant=debian10 --ram=$((1*1024)) --vcpus=1 --disk=path=/var/lib/libvirt/images/example-vm.qcow2,bus=virtio,size=5 --graphics=none --check=all=off --extra-args="console=ttyS0" --location=debian-10.7.0-amd64-netinst.iso`
+    - The disk path should match a storage pool path.
+    - Show available OS variants: `osinfo-query os`
+    - This will automatically open a console for the VM. Specify `--noautoconsole` to avoid that.
+    - If it "can't find the kernel" when using `--location`, use `--cdrom` instead. This disallows using arguments like `--extra-args`, so you'll need to find another console.
+    - Make sure the ISO is readable by the QEMU user.
+    - To install using a VNC screen instead of console (e.g. if you need richer graphics or when using `--cdrom` and no console is allocated), replace `--graphics=none` with `--noautoconsole --graphics=vnc,password=<password>`. It only binds to localhost by default, so use something like SSH port forwarding (`ssh -L 5900:127.0.0.1:5900 <user>@<addr>`) to access it remotely.
+- Clone VM:
+    1. Create a source/template VM and make sure it's not running.
+    1. Clone it: `virt-clone --original=<source-vm> --name=<vm> -f <vm>.qcow2`
+- Remove VM: `virsh undefine <vm>`
+- Set VM to automatically start (or disable it): `virsh autostart [--disable] <vm>`
+- Edit VM config: `virsh edit <vm>`
+- Show VM config: `virsh dumpxml <vm>`
+- Show VM graphics URI: `virsh domdisplay <vm>`
+    - For VNC, the shown port is offset from port 5900.
+- Run QEMU monitor command: `qemu-monitor-command <vm> --hmp <command>`
+
+### Networking
+
+- Set up networking on host:
+    1. Enable IP forwarding on the system (IPv4 and IPv6).
+    1. Create bridges to connect VMs to networks.
+    1. Add firewall rules to allow traffic.
+- Show networks: `virsh net-list`
+- Show network config: `virsh net-dumpxml <network>`
+- Edit network config (without applying it): `virsh net-edit <network>`
+- Apply changed network config: Restart libvirt or reboot the system.
+- The default network interface is `virbr0`, called `default` in libvirt.
+- Enable the default network: `virsh net-start default && virsh net-autostart default`
+
+### Storage
+
+- Effectively, the pool is a directory while the files in it (typically disk images) are the volumes.
+- Pool basics:
+    - Show pools: `virsh pool-list [--all] [--details]`
+- Volume basics:
+    - Show volumes: `virsh list-vol <pool> [--details]`
+    - Show volume info: `virsh vol-info <file>`
+- Create default storage pool:
+    1. Create it: `virsh pool-define-as default --type=dir --target=/var/lib/libvirt/images`
+    1. Start it: `virsh pool-start default`
+    1. Autostart it: `virsh pool-autostart default`
+- Resize disk: `qemu-img resize <file> <size-change>` (e.g. +1G)
+- Cold backup of VM:
+    - Make sure the VM is stopped so that the disk image is consistent.
+    - Backup the image in `/var/lib/libvirt/images/`.
+    - Backup the config in `/etc/libvirt/qemu/`.
+
+### Snapshots
+
+- The current snapshot means the last one.
+- List snapshots for a VM: `virsh snapshot-list <vm>`
+- Show snapshot info: `virsh snapshot-info --domain=<vm> [--current]`
+- Create snapshot: `virsh snapshot-create-as --domain=<vm> --name=<snapshot> --description=<description>`
+- Revert a snapshot:
+    1. Stop the VM.
+    1. Revert a named or the current snapshot: `virsh snapshot-revert {--domain=<vm> | --current} --snapshotname=<snapshot> [--running]`
+
+### Tuning
+
+Assign more CPU cores. When adding many, attempt to assign every hyperthreaded twin to the same VM.
+TODO Cache mode write-through and write-back.
+Consider enabling huge pages.
+Consider using memory ballooning to save memory on the host when the VM doesn't need it.
+
+### Miscellanea
+
+- Use `LIBVIRT_DEFAULT_URI=qemu:///system` to use the system URI.
+- To repair a corrupted QEMU disk, try using `guestfish`.
+
+{% include footer.md %}

config/virt-cont/proxmox-ve.md

@@ -6,10 +6,7 @@ breadcrumbs:
 ---
 {% include header.md %}
 
-### Using
-{:.no_toc}
-
-- Proxmox VE 6
+Using **Proxmox VE 6**.
 
 ## Host
 

index.md

@@ -99,6 +99,7 @@ Random collection of config notes and miscellaneous stuff. _Technically not a wi
 ### Virtualization & Containerization
 
 - [Docker](config/virt-cont/docker/)
+- [libvirt & KVM](config/virt-cont/libvirt-kvm/)
 - [Proxmox VE](config/virt-cont/proxmox-ve/)
 
 ## Information Technology