Installation

Historical documentation restored from the Pigsty v1.5.1 tag.

It takes 3 steps to install Pigsty: Prepare, Configure, Execute Playbook

There are two typical modes: Singleton & Cluster Management

Singleton Meta: Install pigsty on one single meta node. (for PostgreSQL dev/test, RDS monitoring, etc…)
Cluster Manage: Initiate control from meta node, and manage & monitoring more nodes & databases.

Singleton Meta

When Pigsty is installed on one single node. It will init pigsty infra with a battery including Postgres as CMDB. You may use that database directly.

TL; DR

Prepare a new node : Linux x86_64 CentOS 7.8.2003, with root or sudo access, then download pigsty

curl -SL https://github.com/Vonng/pigsty/releases/download/v1.5.1/pigsty.tgz | gzip -d | tar -xC ~  # Download Latest Pigsty Source
cd ~/pigsty; ./configure                               # Configure pigsty with ip & template
./infra.yml                                            # Install pigsty on current meta node

If you have available Mac/PC/Laptop/Cloud Account, Sandbox could be handy to create vm for you.

After installation. port 5432 is ready for postgres database connection @ postgres://dbuser_dba:[email protected]:5432/meta.

And a Nginx is serving & proxy all WebUI services @ port 80.

Download Directly via Curl

curl https://github.com/Vonng/pigsty/releases/download/v1.5.1/pigsty.tgz -o /tmp/pigsty.tgz
curl https://github.com/Vonng/pigsty/releases/download/v1.5.1/pkg.tgz    -o /tmp/pkg.tgz

Access Web Services

Nginx port 80 is the default access endpoint for all Web GUI services. Although it is possible to bypass Nginx add access services directly by port, such as Grafana on 3000, it is highly recommended to use domain names. If you don’t have public domain names, considering use Static DNS on your localhost with make dns.

http://g.pigsty or http://<primary_ip>:3000 is the portal of the monitoring system (username: admin, password: pigsty)

Visit http://<primary_ip>:3000 / http://g.pigsty to visit Pigsty Grafana (username: admin, password: pigsty)

Cluster Manage

Pigsty can also be used as a controller & monitoring infrastructure for production-grade large-scale nodes & databases.

You can initiate control from meta node, and add more nodes into Pigsty.

You can deploy different types of High available PostgreSQL Clusters, Redis clusters, or Greenplum/MatrixDB data warehouse.

./nodes.yml  -l pg-test      # init 3 nodes of cluster pg-test
./pgsql.yml  -l pg-test      # init HA PGSQL Cluster pg-test
./redis.yml  -l redis-test   # init redis cluster redis-test
./pgsql-matrixdb.yml -l mx-*  # init MatrixDB: mx-mdw,mx-sdw

Sandbox

There’s a standard 4-node demo env for pigsty, named ‘Sandbox’. You can create a sandbox on your laptop with Vagrant, or create them on public cloud with Terraform. The same deployment procedure could apply to production deployment.

Take sandbox as an example, install pigsty on meta node 10.10.10.10 with:

./infra.yml     # install pigsty on meta node 10.10.10.10

Nodes Init

Now add 3 more nodes 10.10.10.11, 10.10.10.12, 10.10.10.13 into pigsty nodes.yml

./nodes.yml -l pg-test        # init 3 nodes of cluster pg-test

After that, these 3 nodes are set with DCS service, node exporter & promtail logger. And can be used for subsequent database deployment. check Config: Nodes & Playbook: Nodes for details.

PostgreSQL Deploy

Create a classic 1 primary & 2 replicas HA PostgreSQL cluster with pgsql.yml playbook.

./pgsql.yml -l pg-test  # init pgsql cluster pg-test

Once complete, you can check the newly created PostgreSQL cluster from Monitoring Dashboards.

Check Config: PGSQL, Customize: PGSQL, and Playbook: PGSQL for more details.

Redis Deployment

In addition to the PostgreSQL cluster, you can deploy various other types of clusters and even other types of databases.

For example, to deploy Redis:

./configure -m redis    # use redis config template
./nodes.yml             # init nodes for redis cluster
./redis.yml             # create redis on those nodes

Check Config: REDIS and Playbook: REDIS for more details.

MatrixDB Deployment

To deploy the open-source time-series data warehouse MatrixDB (Greenplum7):

./configure -m mxdb   # Use the sandbox environment MatrixDB configuration file template
./download matrix     # Download the MatrixDB package and build the local source
./infra.yml -e no_cmdb=true # If meta node is part of it, add no_cmdb var to skip CMDB creation
./nodes.yml           # configure all nodes for MatrixDB installation
./pgsql-matrixdb.yml   # Install MatrixDB on those nodes