Skip to content

Latest commit

 

History

History
135 lines (97 loc) · 6.1 KB

README.md

File metadata and controls

135 lines (97 loc) · 6.1 KB
Raw

Stacki Cluster Up

This project will bring up a cluster of boxes consisting of a Stacki frontend and an optional number of backends.

Installation

You need to have Vagrant installed, and either VirtualBox or Libvirt/KVM. If using Libvirt/KVM, you also need to install the vagrant-libvirt plugin and NFS server on the host. Then clone this project on your computer.

Also of note: You need to be on the Teradata network, either in the building or VPN'ed in, to use a SLES version of Stacki to bring up a cluster with this tool.

Usage

Open up a terminal on your computer, change to the project directory, then run:

./cluster-up.sh [options...] STACKI_ISO

where STACKI_ISO is the path to the Stacki ISO you want to base the cluster on. It can be either a SLES 12 or Centos 7 based ISO, and the path can be relative or absolute. The ISO argument can also be a URL, which will be downloaded into the download directory before bringing up the cluster.

There are some optional arguments you can pass to the command:

--backends=[0-253]            The number backends to create. Default: 0
--fqdn=FQDN                   The FQDN of the frontend. Default: cluster-up-frontend.localdomain
--name=NAME                   The name to uniquely identify this cluster. Default: cluster-up-RRRRRRRR
--download-dir=DIRECTORY      The directory to store installer ISOs. Default: '.'
--use-the-src-dir=DIRECTORY   The directory will be mounted and symlinked into the frontend.
--export-frontend             Export the frontend as a vagrant box with the same name as the STACK_ISO
--forward-ports=SRC:DST[,...] Comma separated list of ports to forward.

You can also specify the following flags, which will switch cluster-up to create the second interface to a bridge interface on the VM host. If you specify a bridged interface, then no backends will be automatically created, as they rely on host only networking. Instead, you can create your own backends over the bridged interface.

--bridge=INTERFACE            Bridge interface on the host. Default: Use host only networking
--ip=IP_ADDRESS               IP Address for the bridge network. Default: Use DHCP
--netmask=NETMASK             The netmask for the bridge network. Default: 255.255.255.0
--gateway=GATEWAY_ADDRESS     Gateway address for the bridged interface.
--dns=DNS_ADDRESS[,DNS2]      Comma separated list of DNS servers.

The script will create a generic OS frontend box in Virtualbox or KVM, install Stacki on it, then pxe-boot the number of backends requested (None by default) and install an OS on them from the Stacki frontend.

Once all of that is done, you can ssh into the frontend as root with the command:

vagrant ssh frontend

The backends are named 'backend-0-N' where N is the number of backends created, starting at zero. So, to ssh into the backend-0-0 you would run:

vagrant ssh backend-0-0

When you are done with the cluster, you can destroy it with the command:

./cluster-down.sh

Examples

The simplest invocation of the tool is to create a cluster with the frontend and no backends. That command would simply look like:

./cluster-up.sh stacki-05.02.04.00-redhat7.x86_64.disk1.iso

If you want a handful of backends, assign your frontend a specific FQDN, and to fetch the ISO from a URL, the comand would look like:

./cluster-up.sh \
    --backends=3 \
    --fqdn=stacki-frontend.example.com \
    http://example.com/stacki-05.02.04.00-redhat7.x86_64.disk1.iso

You can create a development environment, with a checkout of the Stacki code repository mapped into it, by using the --use-the-src-dir flag like so:

./cluster-up.sh --use-the-src-dir=~/Code/stacki \
    http://example.com/stacki-05.02.04.00-redhat7.x86_64.disk1.iso

A more complicated use of the tool is to create a Stacki frontend that can provision physical backends. For this, you need to have a bridged network interface set up in either KVM or Virtualbox.

You then specify all your bridge network infomation via flags, such as:

./cluster-up.sh \
    --fqdn=sd-stacki-147.labs.teradata.com \
    --bridge=br1 \
    --ip=10.25.250.147 \
    --gateway=10.25.250.254 \
    --dns=153.64.251.200 \
    stacki-05.02.04.00-sles12.x86_64.disk1.iso

It is a little annoying that in Virtualbox, at least under the Mac, the bridge interface names are a little unwieldy, and you have to specify the entire name, or else it will prompt you several times to select your bridge.

So, under Virtualbox, your command might look like:

./cluster-up.sh \
    --fqdn=sd-stacki-147.labs.teradata.com \
    --bridge="en8: Belkin USB-C LAN" \
    --ip=10.25.250.147 \
    --gateway=10.25.250.254 \
    --dns=153.64.251.200 \
    stacki-05.02.04.00-sles12.x86_64.disk1.iso

Hooks

Customization scripts can be added into a set of folders under the hooks folder, to be invoked at various points in the cluster-up.sh process. The hook scripts must have the execute bit set.

The various hook run points are:

  • pre-cluster-up - After parsing the command arguments but prior to anything else.

  • pre-frontend - Right before vagrant is used to start the frontend VM.

  • post-frontend - After the installation reboot for the frontend.

  • pre-export - Before the vagrant box is exported but after the frontend has been set up to install backends. Only ran if the --export-frontend flag is set.

  • post-export - After the vagrant box has been exported and the frontend has been restarted. Only ran if the --export-frontend flag is set.

  • pre-backends - After discovery has been started on the frontend but before any backends have been started by vagrant. Only ran if backends are to be provisioned.

  • post-backends - After all backends have been provisioned but before discovery has been disabled on the frontend. Only ran if backends are to be provisioned.

  • post-cluster-up - Right before the cluster-up.sh exits and after the success message has been printed out to the terminal.