Scripts for working with and optimizing proxmox, combining all xshok-proxmox forks together into a single repo and further extend and improve (incl. documentation).
Use at your own risk.
Maintained and provided by Melroy van den Berg.
Note
Pull requests are very much welcome!
Turns a fresh proxmox install into an optimised proxmox host.
(not required if server is already setup with the installimage-proxmox.sh script)
reboot-quickalias command which uses kexec to load and boot the latest kernel, without the need for a hardware reboot.- Disable the enterprise repo, enable the public repo, Add non-free sources
- Fixes known bugs (public key missing, max user watches, etc)
- Update the system
- Detect AMD EPYC CPU and Apply Fixes
- Force APT to use IPv4
- Update proxmox and install various system utils
- Customise bashrc
- Add the latest ceph provided by proxmox
- Disable portmapper / rpcbind (security)
- Ensure Entropy Pools are Populated, prevents slowdowns whilst waiting for entropy
- Protect the web interface with fail2ban
- Detect if is running in a virtual machine and install the relavant guest agent
- Install ifupdown2 for a virtual internal network allows rebootless networking changes (not compatible with openvswitch-switch)
- Limit the size and optimise journald
- Install kernel source headers
- Install kexec, allows for quick reboots into the latest updated kernel set as primary in the boot-loader.
- Ensure ksmtuned (ksm-control-daemon) is enabled and optimise according to ram size
- Set language, if chnaged will disable XS_NOAPTLANG
- Increase max user watches, FD limit, FD ulimit, max key limit, ulimits
- Optimise logrotate
- Lynis security scan tool by Cisofy
- Increase Max FS open files
- Optimise Memory
- Pretty MOTD BANNER
- Enable Network optimising
- Save bandwidth and skip downloading additional languages, requires XS_LANG="en_US.UTF-8"
- Disable enterprise proxmox repo
- Remove subscription banner
- Install Open vSwitch for a virtual internal network
- Detect if this is an OVH server and install OVH Real Time Monitoring
- Set pigz to replace gzip, 2x faster gzip compression
- Bugfix: high swap usage with low memory usage
- Enable TCP BBR congestion control
- Enable TCP Fast Open (TFO)
- Enable testing proxmox repo
- Automatically Synchronize the time
- Set Timezone, empty = set automatically by IP
- Install common system utilities
- Increase vzdump backup speed
- Optimise ZFS arc size accoring to memory size
- Install zfs-auto-snapshot
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/install-post.sh -c -O install-post.sh && \
bash install-post.sh && \
rm install-post.shScript should return 0 (zero) if successful.
User defined options for (install-post.sh) post-installation script for Proxmox can be set in the xs-install-post.env file, see the sample: xs-install-post.env.sample.
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/xs-install-post.env.sample -c -O xs-install-post.env
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/install-post.sh -c -O install-post.sh
nano xs-install-post.env
bash install-post.shExample to enable the MOTD banner
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/install-post.sh -c -O install-post.sh
export XS_MOTD="yes"
bash install-post.shRecommeneded partitioning scheme:
- Raid 1 (mirror) 40.000MB ext4 /
- Raid 1 (mirror) 30.000MB ext4 /xshok/zfs-cache only create if an ssd and there is 1+ unused hdd which will be made into a zfspool
- Raid 1 (mirror) 5.000MB ext4 /xshok/zfs-slog only create if an ssd and there is 1+ unused hdd which will be made into a zfspool
- SWAP
- SSD less than 130GB = 16GB swap
- SSD more than 130GB and RAM less than 64GB = 32GB swap
- SSD more than 130GB and RAM more than 64GB = 64GB swap
- Remaining for lv xfs /var/lib/vz (LVM)
See ovh directory
curl -O https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/debian-2-proxmox/debian12-2-proxmox8.sh && chmod +x debian12-2-proxmox8.sh
./debian12-2-proxmox8.shAssumptions: Debian11 installed with a valid FQDN hostname set
- Tested on KVM, VirtualBox and Dedicated Server
- Will automatically detect cloud-init and disable.
- Will automatically generate a correct /etc/hosts
- Note: will automatically run the install-post.sh script
curl -O https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/debian-2-proxmox/debian11-2-proxmox7.sh && chmod +x debian11-2-proxmox7.sh
./debian11-2-proxmox7.shcurl -O https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/debian-2-proxmox/debian10-2-proxmox6.sh && chmod +x debian10-2-proxmox6.sh
./debian10-2-proxmox6.shcurl -O https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/debian-2-proxmox/debian9-2-proxmox5.sh && chmod +x debian9-2-proxmox5.sh
./debian9-2-proxmox5.shThere can be security implications as the LXC container is running in a higher privileged mode.
curl https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/helpers/pve-enable-lxc-docker.sh --output /usr/sbin/pve-enable-lxc-docker && chmod +x /usr/sbin/pve-enable-lxc-docker
pve-enable-lxc-docker container_idConverts the a MDADM BASED LVM into a ZFS raid 1 (mirror)
- Defaults to mount point: /var/lib/vz
- Optional: specify the LVM_MOUNT_POINT ( ./lvm-2-zfs.sh LVM_MOUNT_POINT )
- Creates the following storage/rpools
- zfsbackup (rpool/backup)
- zfsvmdata (rpool/vmdata)
- /var/lib/vz/tmp_backup (rpool/tmp_backup)
- Will automatically detect the required raid level and optimise.
- 1 Drive = zfs
- 2 Drives = mirror
- 3-5 Drives = raidz-1
- 6-11 Drives = raidz-2
- 11+ Drives = raidz-3
Warning
WILL DESTROY ALL DATA ON LVM_MOUNT_POINT !!!
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/zfs/lvm-2-zfs.sh -c -O lvm-2-zfs.sh && chmod +x lvm-2-zfs.sh
./lvm-2-zfs.shCreates a zfs pool from specified devices
- Will automatically detect the required raid level and optimise
- 1 Drive = zfs (single)
- 2 Drives = mirror (raid1)
- 3-5 Drives = raidz-1 (raid5)
- 6-11 Drives = raidz-2 (raid6)
- 11+ Drives = raidz-3 (raid7)
Warning
WILL DESTROY ALL DATA ON SPECIFIED DEVICES!!!
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/zfs/createzfs.sh -c -O createzfs.sh && chmod +x createzfs.sh
./createzfs.sh poolname /dev/device1 /dev/device2Create ZFS cache and slog from /xshok/zfs-cache and /xshok/zfs-slog partitions and adds them to a zpool optional
Creates a zfs pool from specified devices
- Will automatically mirror the slog and stripe the cache if there are multiple drives
Warning
WILL DESTROY ALL DATA ON SPECIFIED PARTITIONS!!!
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/zfs/xshok_slog_cache-2-zfs.sh -c -O xshok_slog_cache-2-zfs.sh && chmod +x xshok_slog_cache-2-zfs.sh
./xshok_slog_cache-2-zfs.sh poolnameAutodetects the correct settings (interface, gatewat, netmask, etc) Supports IPv4 and IPv6, Private Network uses 10.10.10.1/24 Also installs and properly configures the isc-dhcp-server to allow for DHCP on the vmbr1 (NAT)
ROUTED (vmbr0):
- All traffic is routed via the main IP address and uses the MAC address of the physical interface.
- VM's can have multiple IP addresses and they do NOT require a MAC to be set for the IP via service provider
NAT (vmbr1):
- Allows a VM to have internet connectivity without requiring its own IP address
- Assignes 10.10.10.100 - 10.10.10.200 via DHCP
Public IP's can be assigned via DHCP, adding a host define to the /etc/dhcp/hosts.public file
Tested on OVH and Hetzner based servers
ALSO CREATES A NAT Private Network as vmbr1.
Note
WILL OVERWRITE /etc/network/interfaces.
A backup will be created as /etc/network/interfaces.timestamp
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/networking/network-configure.sh -c -O network-configure.sh && chmod +x network-configure.sh
./network-configure.sh && rm network-configure.shIf no interface is specified the default gateway interface will be detected and used.
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/networking/network-addiprange.sh -c -O network-addiprange.sh && chmod +x network-addiprange.sh
./network-addiprange.sh ip.xx.xx.xx/cidr interface_optionaltinc private mesh vpn/network which supports multicast, ideal for private cluster communication
wget https://raw.githubusercontent.com/melroy89/xshok-proxmox/master/networking/tincvpn.sh -c -O tincvpn.sh && chmod +x tincvpn.sh
./tincvpn.sh -hcat /etc/hosts
# global ips for tinc servers
11.11.11.11 host1
22.22.22.22 host2
33.33.33.33 host3bash tincvpn.sh -i 1 -c host2bash tincvpn.sh -i 2 -c host3bash tincvpn.sh -i 3 -c host1Run the following on the guest alpine linux
apk update && apk add qemu-guest-agent acpi
echo 'GA_PATH="/dev/vport2p1"' >> /etc/conf.d/qemu-guest-agent
rc-update add qemu-guest-agent default
rc-update add acpid default
/etc/init.d/qemu-guest-agent restartRun the following on the proxmox server, ensure you have a valid DNS for the server which resolves
pvenode acme account register default mail@example.invalid
pvenode config set --acme domains=example.invalid
pvenode acme cert order# list all snapshots
zfs list -t snapshot
# create a pre-rollback snapshot
zfs-auto-snapshot --verbose --label=prerollback -r //
# rollback to a specific snapshot
zfs rollback <snapshotname>