Bootstrap high available PostgreSQL instance using a shared storage

This Vagrantfile bootstraps a fresh virtualized cluster with:

• minimum two VM under Debian
• watchdog enabled on all nodes
• one PostgreSQL instance
• one shared storage for PostgreSQL data and Storage Based Death
• one vIP moving with the PostgreSQL instance
• one dummy service avoiding to colocate with PostgreSQL, just for fun and demo

This has been developed and tested using:

• vagrant 2.2.14
• vagrant-libvirt 0.5.3
• Debian 11 bullseye as VM OS (vagrant box debian/bullseye64)
• Pacemaker 2.0.5
• pcs 0.10.8

Prerequisites:

You need vagrant and vagrant-libvirt.

apt install make vagrant vagrant-libvirt libvirt-clients # for Debian-like
yum install make vagrant vagrant-libvirt libvirt-client # for RH-like
dnf install make vagrant vagrant-libvirt libvirt-client # for recent RH-like
systemctl enable --now libvirtd

Alternatively, you can install vagrant-libvirt using:

apt install libvirt-dev ruby-dev
vagrant plugin install vagrant-libvirt

Creating the cluster

To create the cluster, run:

cd vagrant-patroni
make all

After some minutes and tons of log messages, you can connect to your servers using eg.:

vagrant ssh srv1

Play with the cluster!

srv1> sudo -i

srv1# pcs resource

srv1# crm_mon -1D

srv1# crm_mon -1Dn

srv1# # move the resource away (and clear the temp constraint)
srv1# pcs resource move pgsqlgroup --wait
srv1# pcs resource clear pgsqlgroup --wait

srv1# # Kill the node hosting pgsql.
srv1# # Open crm_mon in another term to observe the reaction
srv1# pcs stonith fence srv2
srv1# crm_mon -frAno

srv1# # after the reboot of srv2 node, start the cluster on the node
srv1# pcs cluster start srv2

Customizing the cluster

Copy vagrant.yml-dist as vagrant.yml read comments and edit as needed.

Destroying the cluster

To destroy your cluster, run:

make clean

Tips

Find all existing VM created by vagrant on your system:

vagrant global-status