Module: INFRA

• 1: Configuration
• 2: Parameters
• 3: Playbook
• 4: Monitoring
• 5: Metrics
• 6: FAQ
• 7: Administration
  • 7.1: Nginx Management
  • 7.2: Software Repository
  • 7.3: Domain Management
  • 7.4: Module Management
  • 7.5: CA and Certificates
  • 7.6: Grafana High Availability: Using PostgreSQL Backend

infra_portal:  # domain names and upstream servers
  home         : { domain: i.pigsty }
  grafana      : { domain: g.pigsty ,endpoint: "${admin_ip}:3000" , websocket: true }
  prometheus   : { domain: p.pigsty ,endpoint: "${admin_ip}:8428" }   # VMUI
  alertmanager : { domain: a.pigsty ,endpoint: "${admin_ip}:9059" }
  blackbox     : { endpoint: "${admin_ip}:9115" }
  vmalert      : { endpoint: "${admin_ip}:8880" }
  #logs         : { domain: logs.pigsty ,endpoint: "${admin_ip}:9428" }
  #minio        : { domain: sss.pigsty  ,endpoint: "${admin_ip}:9001" ,scheme: https ,websocket: true }

curl -L http://i.pigsty/pigsty.repo -o /etc/yum.repos.d/pigsty.repo

[pigsty-local]
name=Pigsty local $releasever - $basearch
baseurl=file:///www/pigsty/
enabled=1
gpgcheck=0

# Configure single INFRA node
infra: { hosts: { 10.10.10.10: { infra_seq: 1 } }}

# Configure two INFRA nodes
infra:
  hosts:
    10.10.10.10: { infra_seq: 1 }
    10.10.10.11: { infra_seq: 2 }

./infra.yml     # Install INFRA module on infra group
./infra-rm.yml  # Full removal, including data and packages

./infra.yml -t repo              # Create local repo from internet or offline package

./infra.yml -t repo_dir          # Create local repo directory
./infra.yml -t repo_check        # Check if local repo already exists
./infra.yml -t repo_prepare      # If exists, use existing local repo
./infra.yml -t repo_build        # If not exists, build local repo from upstream
./infra.yml     -t repo_upstream     # Handle upstream repo files in /etc/yum.repos.d
./infra.yml     -t repo_remove       # If repo_remove == true, delete existing repo files
./infra.yml     -t repo_add          # Add upstream repo files to /etc/yum.repos.d (or /etc/apt/sources.list.d)
./infra.yml     -t repo_url_pkg      # Download packages from internet defined by repo_url_packages
./infra.yml     -t repo_cache        # Create upstream repo metadata cache with yum makecache / apt update
./infra.yml     -t repo_boot_pkg     # Install SOW and dnf/yum download utilities
./infra.yml     -t repo_pkg          # Download packages & dependencies from upstream repos
./infra.yml     -t repo_create       # Atomically create RPM/APT metadata with sow create --pigsty
./infra.yml     -t repo_use          # Add newly built repo to /etc/yum.repos.d | /etc/apt/sources.list.d
./infra.yml -t repo_nginx        # If no nginx serving, start nginx as web server

./infra.yml     -t repo_upstream     # Add upstream repos defined in repo_upstream to INFRA nodes
./infra.yml     -t repo_pkg          # Download packages and dependencies from upstream repos
./infra.yml     -t repo_create       # Create/update local RPM/APT repository metadata with SOW

./infra.yml -t infra           # Configure infrastructure
./infra.yml -t infra_user      # Setup infra OS user group
./infra.yml -t infra_dir       # Create infrastructure data, config, and runtime directories
./infra.yml -t infra_env       # Configure admin environment: env_patroni, env_pg, env_pgadmin, env_etcd, env_pglog, env_var
./infra.yml -t infra_pkg       # Install software packages required by INFRA: infra_packages
./infra.yml -t infra_cert      # Issue certificates for infra components
./infra.yml -t dns             # Configure DNSMasq: dns_config, dns_record, dns_launch
./infra.yml -t nginx           # Configure Nginx: nginx_config, nginx_cert, nginx_static, nginx_launch, nginx_certbot, nginx_reload, nginx_exporter
./infra.yml -t victoria        # Configure VictoriaMetrics/Logs/Traces: vmetrics|vlogs|vtraces|vmalert
./infra.yml -t alertmanager    # Configure AlertManager: alertmanager_config, alertmanager_launch
./infra.yml -t blackbox        # Configure Blackbox Exporter: blackbox_config, blackbox_launch
./infra.yml -t grafana         # Configure Grafana: grafana_clean, grafana_config, grafana_plugin, grafana_launch, grafana_provision
./infra.yml -t infra_register  # Register infra components to VictoriaMetrics / Grafana

./infra.yml -t nginx_index                        # Re-render Nginx homepage content
./infra.yml -t nginx_config,nginx_reload          # Re-render Nginx portal config, expose new upstream services
./infra.yml -t vmetrics_config,vmetrics_launch    # Regenerate VictoriaMetrics main config and restart service
./infra.yml -t vlogs_config,vlogs_launch          # Re-render VictoriaLogs config
./infra.yml -t vmetrics_clean                     # Clean VictoriaMetrics storage data directory
./infra.yml -t grafana_plugin                     # Download Grafana plugins from internet

./infra-rm.yml               # Full removal: deregister, stop, remove config/environment/data, and uninstall packages
./infra-rm.yml -t deregister # Only deregister monitoring targets, datasources, and log collection
./infra-rm.yml -t service    # Stop infrastructure services on INFRA
./infra-rm.yml -t data       # Remove INFRA data
./infra-rm.yml -t package    # Uninstall packages

1 - Configuration

Configuration Guide

INFRA = primarily monitoring infrastructure, optional for PostgreSQL databases.

Unless manually configured to depend on DNS/NTP services on INFRA nodes, INFRA module failures typically don’t affect PG cluster operations.

Single INFRA node suffices for most scenarios. Prod env recommends 2-3 INFRA nodes for HA.

For better resource utilization, ETCD module (required by PG HA) can share nodes with INFRA module.

Using more than 3 INFRA nodes provides little additional benefit, but more ETCD nodes (e.g., 5) can improve DCS availability.

Configuration Examples

Add node IPs to infra group in config inventory, assign INFRA instance number infra_seq.

Default single INFRA node config:

all:
  children:
    infra: { hosts: { 10.10.10.10: { infra_seq: 1 } }}

By default, 10.10.10.10 placeholder replaced with current node’s primary IP during config.

Use infra.yml playbook to init INFRA module on nodes.

More Nodes

Two INFRA nodes config:

all:
  children:
    infra:
      hosts:
        10.10.10.10: { infra_seq: 1 }
        10.10.10.11: { infra_seq: 2 }

Three INFRA nodes config (with params):

all:
  children:
    infra:
      hosts:
        10.10.10.10: { infra_seq: 1 }
        10.10.10.11: { infra_seq: 2, repo_enabled: false }
        10.10.10.12: { infra_seq: 3, repo_enabled: false }
      vars:
        grafana_clean: false
        vmetrics_clean: false
        vlogs_clean: false
        vtraces_clean: false

INFRA High Availability

Most INFRA module components = “stateless/identical state”. For HA, focus on “load balancing”.

HA achievable via Keepalived L2 VIP or HAProxy L4 load balancing. L2 VIP recommended for L2-reachable networks.

Config example:

infra:
  hosts:
    10.10.10.10: { infra_seq: 1 }
    10.10.10.11: { infra_seq: 2 }
    10.10.10.12: { infra_seq: 3 }
  vars:
    vip_enabled: true
    vip_vrid: 128
    vip_address: 10.10.10.8
    vip_interface: eth1

    infra_portal:
      home         : { domain: i.pigsty }
      grafana      : { domain: g.pigsty ,endpoint: "10.10.10.8:3000" , websocket: true }
      prometheus   : { domain: p.pigsty ,endpoint: "10.10.10.8:8428" }
      alertmanager : { domain: a.pigsty ,endpoint: "10.10.10.8:9059" }
      blackbox     : { endpoint: "10.10.10.8:9115" }
      vmalert      : { endpoint: "10.10.10.8:8880" }

Set VIP-related params and modify service endpoints in infra_portal.

Nginx Configuration

See Nginx Parameter Config and Tutorial: Nginx.

Local Repo Configuration

See Repo Parameter Config.

DNS Configuration

See DNS Parameter Config and Tutorial: DNS.

NTP Configuration

See NTP Parameter Config.

2 - Parameters

The INFRA module is responsible for deploying Pigsty’s infrastructure components: local software repository, Nginx, DNSMasq, VictoriaMetrics, VictoriaLogs, Grafana, Alertmanager, Blackbox Exporter, and other monitoring and alerting infrastructure.

Pigsty v4.x uses VictoriaMetrics to replace Prometheus and VictoriaLogs to replace Loki, providing a superior observability solution.

Parameter Overview

META parameters define Pigsty metadata, including version string, admin node IP, repository mirror region, default language, and proxy settings.

CA parameters configure Pigsty’s self-signed CA certificate management, including CA creation, CA name, and certificate validity.

INFRA_ID parameters define infrastructure node identity, including node sequence number, service portal configuration, and data directory.

REPO parameters configure the local software repository, including repository enable switch, directory paths, upstream source definitions, and packages to download.

INFRA_PACKAGE parameters define RPM/DEB packages to install on infrastructure nodes.

NGINX parameters configure Nginx web server and reverse proxy, including enable switch, ports, SSL mode, certificates, and basic authentication.

DNS parameters configure DNSMasq DNS server, including enable switch, listen port, and dynamic DNS records.

VICTORIA parameters configure the VictoriaMetrics/Logs/Traces observability stack, including enable switches, ports, and data retention policies.

PROMETHEUS parameters configure Alertmanager and Blackbox Exporter, providing alert management and network probing capabilities.

GRAFANA parameters configure the Grafana visualization platform, including enable switch, port, admin credentials, and data source configuration.

META

This section defines Pigsty deployment metadata: version string, admin node IP address, repository mirror region, default language, and HTTP(S) proxy for downloading packages.

version: v4.5.0                   # pigsty version string
admin_ip: 10.10.10.10             # admin node ip address
region: default                   # upstream mirror region: default,china,europe
language: en                      # default language: en or zh
proxy_env:                        # global proxy env when downloading packages
  no_proxy: "localhost,127.0.0.1,10.0.0.0/8,192.168.0.0/16,*.pigsty,*.aliyun.com,mirrors.*,*.myqcloud.com,*.tsinghua.edu.cn"
  # http_proxy:  # set your proxy here: e.g http://user:pass@proxy.xxx.com
  # https_proxy: # set your proxy here: e.g http://user:pass@proxy.xxx.com
  # all_proxy:   # set your proxy here: e.g http://user:pass@proxy.xxx.com

version

name: version, type: string, level: G

Pigsty version string. The current source default is v4.5.0.

Pigsty uses this version string internally for feature control and content rendering. Do not modify this parameter arbitrarily.

Pigsty uses semantic versioning, and the version string typically starts with the character v, e.g., v4.5.0.

admin_ip

name: admin_ip, type: ip, level: G

Admin node IP address, default is the placeholder IP address: 10.10.10.10

The node specified by this parameter will be treated as the admin node, typically pointing to the first node where Pigsty is installed, i.e., the control node.

The default value 10.10.10.10 is a placeholder that will be replaced with the actual admin node IP address during configure.

Many parameters reference this parameter, such as:

• infra_portal
• repo_endpoint
• repo_upstream
• dns_records
• node_default_etc_hosts
• node_etc_hosts

In these parameters, the string ${admin_ip} will be replaced with the actual value of admin_ip. Using this mechanism, you can specify different admin nodes for different nodes.

region

name: region, type: enum, level: G

Upstream mirror region, available options: default, china, europe, default is default

If a region other than default is set, and there’s a corresponding entry in repo_upstream with a matching baseurl, it will be used instead of the default baseurl.

For example, if your region is set to china, Pigsty will attempt to use Chinese mirror sites to accelerate downloads. If an upstream repository doesn’t have a corresponding China region mirror, the default upstream mirror site will be used instead. Additionally, URLs defined in repo_url_packages will be replaced from repo.pigsty.io to repo.pigsty.cc to use domestic mirrors.

language

name: language, type: enum, level: G

Default language setting, options are en (English) or zh (Chinese), default is en.

This parameter affects the language preference of some Pigsty-generated configurations and content, such as the initial language setting of Grafana dashboards.

If you are a Chinese user, it is recommended to set this parameter to zh for a better Chinese support experience.

proxy_env

name: proxy_env, type: dict, level: G

Global proxy environment variables used when downloading packages, default value specifies no_proxy, which is the list of addresses that should not use a proxy:

proxy_env:
  no_proxy: "localhost,127.0.0.1,10.0.0.0/8,192.168.0.0/16,*.pigsty,*.aliyun.com,mirrors.*,*.myqcloud.com,*.tsinghua.edu.cn"
  #http_proxy: 'http://username:password@proxy.address.com'
  #https_proxy: 'http://username:password@proxy.address.com'
  #all_proxy: 'http://username:password@proxy.address.com'

When installing from the Internet in mainland China, certain packages may be blocked. You can use a proxy to solve this problem.

Note that if the Docker module is used, the proxy server configuration here will also be written to the Docker Daemon configuration file.

Note that if the -x parameter is specified during ./configure, the proxy configuration information in the current environment will be automatically filled into the generated pigsty.yaml file.

CA

Pigsty uses self-signed CA certificates to support advanced security features such as HTTPS access, PostgreSQL SSL connections, etc.

ca_create: true                   # allow creation if CA private key is missing? default true
ca_cn: pigsty-ca                  # CA CN name, fixed as pigsty-ca
cert_validity: 7300d              # certificate validity, default 20 years

ca_create

name: ca_create, type: bool, level: G

Allow CA creation when the private key is missing? The default is true.

When set to true, Pigsty creates a new CA private key if files/pki/ca/ca.key is absent. If ca.crt is absent, it uses the existing or newly created private key to issue the CA certificate.

If you already have a CA public-private key pair, you can copy them to the files/pki/ca directory:

• files/pki/ca/ca.crt: CA public key certificate
• files/pki/ca/ca.key: CA private key file

Pigsty reuses an existing CA key pair. If the private key is absent and this parameter is false, deployment stops with an error. If only ca.crt is missing, Pigsty still reissues it with the existing private key. Always back up and restore the matching ca.key and ca.crt together to avoid a certificate/key mismatch.

Be sure to retain and backup the newly generated CA private key file during deployment, as it is crucial for issuing new certificates later.

ca_cn

name: ca_cn, type: string, level: G

CA CN (Common Name), fixed as pigsty-ca, not recommended to modify.

You can use the following command to view the Pigsty CA certificate details on a node:

openssl x509 -text -in /etc/pki/ca.crt

cert_validity

name: cert_validity, type: interval, level: G

Certificate validity period for issued certificates, default is 20 years, sufficient for most scenarios. Default value: 7300d

This parameter affects the validity of all certificates issued by the Pigsty CA, including:

• PostgreSQL server certificates
• Patroni API certificates
• etcd server/client certificates
• Other internal service certificates

Note: The validity of HTTPS certificates used by Nginx is controlled separately by nginx_cert_validity, because modern browsers have stricter requirements for website certificate validity (maximum 397 days).

INFRA_ID

Infrastructure identity and portal definition.

#infra_seq: 1                     # infra node sequence, REQUIRED identity parameter
infra_portal:                     # infrastructure services exposed via Nginx portal
  home : { domain: i.pigsty }     # default home server definition
infra_data: /data/infra           # infrastructure default data directory
infra_services: [...]             # built-in home navigation entries
infra_extra_services: []          # additional home navigation entries

infra_seq

name: infra_seq, type: int, level: I

Infrastructure node sequence number, REQUIRED identity parameter that must be explicitly specified on infrastructure nodes, so no default value is provided.

This parameter is used to uniquely identify each node in multi-infrastructure node deployments, typically using positive integers starting from 1.

Example configuration:

infra:
  hosts:
    10.10.10.10: { infra_seq: 1 }
    10.10.10.11: { infra_seq: 2 }

infra_portal

name: infra_portal, type: dict, level: G

Infrastructure services exposed via Nginx portal. The v4.x default value is very concise:

infra_portal:
  home : { domain: i.pigsty }     # default home server definition

Pigsty will automatically configure the corresponding reverse proxies based on the actually enabled components. Users typically only need to define the home domain name.

Each record consists of a Key and a Value dictionary, where name is the key representing the component name, and the value is an object that can configure the following parameters:

• name: REQUIRED, specifies the name of the Nginx server
  • Default record: home is a fixed name, please do not modify it.
  • Used as part of the Nginx configuration file name, corresponding to: /etc/nginx/conf.d/<name>.conf
  • Nginx servers without a domain field will not generate configuration files but will be used as references.
• domain: OPTIONAL, when the service needs to be exposed via Nginx, this is a REQUIRED field specifying the domain name to use
  • In Pigsty self-signed Nginx HTTPS certificates, the domain will be added to the SAN field of the Nginx SSL certificate
  • Pigsty web page cross-references will use the default domain name here
• endpoint: Usually used as an alternative to path, specifies the upstream server address. Setting endpoint indicates this is a reverse proxy server
  • ${admin_ip} can be used as a placeholder in the configuration and will be dynamically replaced with admin_ip during deployment
  • Default reverse proxy servers use endpoint.conf as the configuration template
  • Reverse proxy servers can also configure websocket and schema parameters
• path: Usually used as an alternative to endpoint, specifies the local file server path. Setting path indicates this is a local web server
  • Local web servers use path.conf as the configuration template
  • Local web servers can also configure the index parameter to enable file index pages
• certbot: Certbot certificate name; if configured, Certbot will be used to apply for certificates
  • If multiple servers specify the same certbot, Pigsty will merge certificate applications; the final certificate name will be this certbot value
• cert: Certificate file path; if configured, will override the default certificate path
• key: Certificate key file path; if configured, will override the default certificate key path
• websocket: Whether to enable WebSocket support
  • Only reverse proxy servers can configure this parameter; if enabled, upstream WebSocket connections will be allowed
• schema: Protocol used by the upstream server; if configured, will override the default protocol
  • Default is http; if configured as https, it will force HTTPS connections to the upstream server
• index: Whether to enable file index pages
  • Only local web servers can configure this parameter; if enabled, autoindex configuration will be enabled to automatically generate directory index pages
• log: Nginx log file path
  • If specified, access logs will be written to this file; otherwise, the default log file will be used based on server type
  • Reverse proxy servers use /var/log/nginx/<name>.log as the default log file path
  • Local web servers use the default Access log
• conf: Nginx configuration file path
  • Explicitly specifies the configuration template file to use, located in roles/infra/templates/nginx or templates/nginx directory
  • If this parameter is not specified, the default configuration template will be used
• config: Nginx configuration code block
  • Configuration text directly injected into the Nginx Server configuration block
• enforce_https: Redirect HTTP to HTTPS
  • Global configuration can be specified via nginx_sslmode: enforce
  • This configuration does not affect the default home server, which will always listen on both ports 80 and 443 to ensure compatibility

infra_data

name: infra_data, type: path, level: G

Infrastructure data directory, default value is /data/infra.

This directory is used to store data files for infrastructure components, including:

• VictoriaMetrics time series database data
• VictoriaLogs log data
• VictoriaTraces trace data
• Other infrastructure component persistent data

It is recommended to place this directory on a separate data disk for easier management and expansion.

infra_services

name: infra_services, type: service[], level: G

Built-in navigation entries on the Pigsty home page. Current defaults include Metrics, Logs, Traces, Monitor Targets, Alert Rules, Alert Manager, CA Certificate, Software Repo, and Explain Visualizer.

Each item can use name, url, desc, and icon, plus the Chinese display fields name_cn and desc_cn. This parameter replaces the entire default list; use infra_extra_services if you only want to append entries.

infra_extra_services

name: infra_extra_services, type: service[], level: G

Navigation entries appended to infra_services. The default is [], and each item uses the same structure as infra_services. For example:

infra_extra_services:
  - { name: My Service, url: 'https://example.com', desc: 'External Service', icon: 'database' }

REPO

This section is about local software repository configuration. Pigsty enables a local software repository (APT/YUM) on infrastructure nodes by default.

During initialization, Pigsty downloads all packages and their dependencies (specified by repo_packages) from the Internet upstream repository (specified by repo_upstream) to {{ nginx_home }} / {{ repo_name }} (default /www/pigsty). The total size of all software and dependencies is approximately 1GB.

The current package candidate is SOW 0.3.0, and the source uses SOW to generate RPM/APT metadata. After a successful build, repo_complete is both a SHA-256 manifest and a completion marker. When it is present, Pigsty skips downloading and rebuilding by default and uses the existing repository. Force a rebuild with ./infra.yml -t repo_build -e repo_build=true.

Both repo_create and cache_create invoke sow create --pigsty directly, with no fallback to createrepo_c or dpkg-scanpackages. If an older offline bundle or local repository does not contain SOW, refresh the media or install SOW from the Pigsty INFRA repository first.

If some packages download too slowly, you can set a download proxy using the proxy_env configuration to complete the initial download, or directly download the pre-packaged offline package, which is essentially a local software repository built on the same operating system.

repo_enabled: true                # create local repo on this infra node?
repo_home: /www                   # repo home directory, default /www
repo_name: pigsty                 # repo name, default pigsty
repo_endpoint: http://${admin_ip}:80 # repo access endpoint
repo_remove: true                 # remove existing upstream repo definitions
repo_modules: infra,node,pgsql    # enabled upstream repo modules
#repo_upstream: []                # upstream repo definitions (inherited from OS variables)
#repo_packages: []                # packages to download (inherited from OS variables)
#repo_extra_packages: []          # extra packages to download
repo_url_packages: []             # extra packages downloaded via URL

repo_enabled

name: repo_enabled, type: bool, level: G/I

Create a local software repository on this infrastructure node? Default is true, meaning all Infra nodes will set up a local software repository.

If you have multiple infrastructure nodes, you can keep only 1-2 nodes as software repositories; other nodes can set this parameter to false to avoid duplicate software download builds.

repo_home

name: repo_home, type: path, level: G

Local software repository home directory, defaults to Nginx’s root directory: /www.

On a fresh installation, if this path does not exist, the role creates a symlink to nginx_data. Existing directories and symlinks are preserved unchanged. Modification is generally discouraged; if required, keep it consistent with nginx_home.

repo_name

name: repo_name, type: string, level: G

Local repository name, default is pigsty. Changing this repository name is not recommended.

The final repository path is {{ repo_home }}/{{ repo_name }}, defaulting to /www/pigsty.

repo_endpoint

name: repo_endpoint, type: url, level: G

Endpoint used by other nodes to access this repository, default value: http://${admin_ip}:80.

Pigsty starts Nginx on infrastructure nodes at ports 80/443 by default, providing local software repository (static files) service.

If you modify nginx_port or nginx_ssl_port, or use a different infrastructure node from the control node, adjust this parameter accordingly.

If you use a domain name, you can add resolution in node_default_etc_hosts, node_etc_hosts, or dns_records.

repo_remove

name: repo_remove, type: bool, level: G/A

Remove existing upstream repository definitions when building the local repository? Default value: true.

When this parameter is enabled, all existing repository files in /etc/yum.repos.d will be moved and backed up to /etc/yum.repos.d/backup. On Debian systems, /etc/apt/sources.list and /etc/apt/sources.list.d are removed and backed up to /etc/apt/backup.

Since existing OS sources have uncontrollable content, using Pigsty-validated upstream software sources can improve the success rate and speed of downloading packages from the Internet.

In certain situations (e.g., your OS is some EL/Deb compatible variant that uses private sources for many packages), you may need to keep existing upstream repository definitions. In such cases, set this parameter to false.

repo_modules

name: repo_modules, type: string, level: G/A

Which upstream repository modules will be added to the local software source, default value: infra,node,pgsql

When Pigsty attempts to add upstream repositories, it filters entries in repo_upstream based on this parameter’s value. Only entries whose module field matches this parameter’s value will be added to the local software source.

During the build stage, infra is automatically added to the effective module list so SOW can be installed. This bootstrap dependency is restored even if a user overrides repo_modules without infra.

Modules are comma-separated. Available module lists can be found in the repo_upstream definitions; common modules include:

• local: Local Pigsty repository
• infra: Infrastructure packages (Nginx, Docker, etc.)
• node: OS base packages
• pgsql: PostgreSQL-related packages
• extra: Extra PostgreSQL extensions
• docker: Docker-related
• redis: Redis-related
• mongo: MongoDB-related
• mysql: MySQL-related
• etc…

repo_upstream

name: repo_upstream, type: upstream[], level: G

Where to download upstream packages when building the local repository? This parameter has no default value. If not explicitly specified by the user in the configuration file, it will be loaded from the repo_upstream_default variable defined in roles/node_id/vars based on the current node’s OS family.

Pigsty provides complete upstream repository definitions for currently supported OS versions (EL 8/9/10, Debian 12/13, Ubuntu 22/24/26), including:

• OS base repositories (BaseOS, AppStream, EPEL, etc.)
• PostgreSQL official PGDG repository
• Pigsty extension repository
• Various third-party software repositories (Docker, Nginx, Grafana, etc.)

Each upstream repository definition contains the following fields:

- name: pigsty-pgsql              # repository name
  description: 'Pigsty PGSQL'     # repository description
  module: pgsql                   # module it belongs to
  releases: [8,9,10]              # supported OS versions
  arch: [x86_64, aarch64]         # supported CPU architectures
  baseurl:                        # repository URL, configured by region
    default: 'https://repo.pigsty.io/yum/pgsql/el$releasever.$basearch'
    china: 'https://repo.pigsty.cc/yum/pgsql/el$releasever.$basearch'
  # meta: { module_hotfixes: 1 } # enable only when explicitly replacing an EL module stream

RPM upstream repositories retain native DNF module filtering by default. Set meta.module_hotfixes only on a repository that must actually replace an EL module stream. Pigsty’s aggregated local repository itself is consumed with module_hotfixes=1, but Pigsty no longer generates fake modules.yaml / ModuleMD metadata.

Users typically don’t need to modify this parameter unless they have special repository requirements. For detailed repository definitions, refer to the configuration files for corresponding operating systems in the roles/node_id/vars/ directory.

repo_packages

name: repo_packages, type: string[], level: G

String array type, where each line is a space-separated list of software packages, specifying packages (and their dependencies) to download using repotrack or apt download.

This parameter has no default value, meaning its default state is undefined. If not explicitly defined, Pigsty will load the default from the repo_packages_default variable defined in roles/node_id/vars:

[ node-bootstrap, infra-package, infra-addons, node-package1, node-package2, node-package3, pgsql-utility, extra-modules ]

Each element in this parameter will be translated according to the package_map in the above files, based on the specific OS distro major version. For example, on EL systems it translates to:

node-bootstrap:          "ansible python3 python3-requests python3-jmespath python3-cryptography dnf-utils sow sshpass"
infra-package:           "nginx dnsmasq etcd haproxy vip-manager node-exporter keepalived-exporter pg-exporter pgbackrest-exporter redis-exporter redis valkey silo mcli sow pig"
infra-addons:            "grafana grafana-plugins grafana-victoriametrics-ds grafana-victorialogs-ds victoria-metrics victoria-logs victoria-traces vlogscli vmutils vector alertmanager"

As a convention, repo_packages typically includes packages unrelated to the PostgreSQL major version (such as Infra, Node, and PGDG Common parts), while PostgreSQL major version-related packages (kernel, extensions) are usually specified in repo_extra_packages to facilitate switching PG major versions.

repo_extra_packages

name: repo_extra_packages, type: string[], level: G/C/I

Used to specify additional packages to download without modifying repo_packages (typically PG major version-related packages), default value is an empty list.

If not explicitly defined, Pigsty will load the default from the repo_extra_packages_default variable defined in roles/node_id/vars:

[ pgsql-main ]

Elements in this parameter undergo package name translation, where $v will be replaced with pg_version, i.e., the current PG major version (default 18).

The pgsql-main here translates on EL systems to:

postgresql$v postgresql$v-server postgresql$v-libs postgresql$v-contrib postgresql$v-plperl postgresql$v-plpython3 postgresql$v-pltcl postgresql$v-llvmjit pg_repack_$v* wal2json_$v* pgvector_$v*

Users can typically specify PostgreSQL major version-related packages here without affecting the other PG version-independent packages defined in repo_packages.

repo_url_packages

name: repo_url_packages, type: object[] | string[], level: G

Packages downloaded directly from the Internet using URLs, default is an empty array: []

You can use URL strings directly as array elements in this parameter, or use object structures to explicitly specify URLs and filenames.

Note that this parameter is affected by the region variable. If you’re in mainland China, Pigsty will automatically replace URLs, changing repo.pigsty.io to repo.pigsty.cc.

INFRA_PACKAGE

These RPM/DEB packages are installed only on INFRA nodes.

infra_packages

name: infra_packages, type: string[], level: G

String array type, where each line is a space-separated list of software packages, specifying packages to install on Infra nodes.

This parameter has no single cross-platform default. If not explicitly specified, Pigsty loads infra_packages_default from the platform-specific file under roles/node_id/vars, based on the operating-system version and CPU architecture.

For example, the current mapping for EL 9 x86_64 is:

infra_packages_default:
  - grafana,grafana-plugins,grafana-victorialogs-ds,grafana-victoriametrics-ds,victoria-metrics,victoria-logs,victoria-traces,vmutils,vlogscli,alertmanager
  - node-exporter,blackbox-exporter,nginx-exporter,pg-exporter,pev2,nginx,dnsmasq,ansible,etcd,python3-requests,redis,mcli,restic,certbot,python3-certbot-nginx

The current mapping for Debian 13 x86_64 is:

infra_packages_default:
  - grafana,grafana-plugins,grafana-victorialogs-ds,grafana-victoriametrics-ds,victoria-metrics,victoria-logs,victoria-traces,vmutils,vlogscli,alertmanager
  - node-exporter,blackbox-exporter,nginx-exporter,pg-exporter,pev2,nginx,dnsmasq,ansible,etcd,python3-requests,redis,mcli,restic,certbot,python3-certbot-nginx

NGINX

Pigsty proxies all web service access through Nginx: Home Page, Grafana, VictoriaMetrics, etc., as well as other optional tools like PGWeb, Jupyter Lab, Pgadmin, Bytebase, and static resources and reports like pev, schemaspy, and pgbadger.

Most importantly, Nginx also serves as the web server for the local software repository (Yum/Apt), used to store and distribute Pigsty packages.

nginx_enabled: true               # enable Nginx on this infra node?
nginx_clean: false                # clean existing Nginx config during init?
nginx_exporter_enabled: true      # enable nginx_exporter?
nginx_exporter_port: 9113         # nginx_exporter listen port
nginx_sslmode: enable             # SSL mode: disable,enable,enforce
nginx_cert_validity: 397d         # self-signed cert validity
nginx_home: /www                  # Nginx content directory (symlink)
nginx_data: /data/nginx           # Nginx actual data directory
nginx_users: {}                   # basic auth users dictionary
nginx_port: 80                    # HTTP port
nginx_ssl_port: 443               # HTTPS port
certbot_sign: false               # sign cert with certbot?
certbot_email: your@email.com     # certbot email
certbot_options: ''               # certbot extra options

nginx_enabled

name: nginx_enabled, type: bool, level: G/I

Enable Nginx on this Infra node? Default value: true.

Nginx is a core component of Pigsty infrastructure, responsible for:

• Providing local software repository service
• Reverse proxying Grafana, VictoriaMetrics, and other web services
• Hosting static files and reports

nginx_clean

name: nginx_clean, type: bool, level: G/A

Clean existing Nginx configuration during initialization? Default value: false.

When set to true, all existing configuration files under /etc/nginx/conf.d/ will be deleted during Nginx initialization, ensuring a clean start.

If you’re deploying for the first time or want to completely rebuild Nginx configuration, you can set this parameter to true.

nginx_exporter_enabled

name: nginx_exporter_enabled, type: bool, level: G/I

Enable nginx_exporter on this infrastructure node? Default value: true.

If this option is disabled, the /nginx health check stub will also be disabled. Consider disabling this when your Nginx version doesn’t support this feature.

nginx_exporter_port

name: nginx_exporter_port, type: port, level: G

nginx_exporter listen port, default value is 9113.

nginx_exporter is used to collect Nginx operational metrics for VictoriaMetrics to scrape and monitor.

nginx_sslmode

name: nginx_sslmode, type: enum, level: G

Nginx SSL operating mode. Three options: disable, enable, enforce, default value is enable, meaning SSL is enabled but not enforced.

• disable: Only listen on the port specified by nginx_port to serve HTTP requests.
• enable: Also listen on the port specified by nginx_ssl_port to serve HTTPS requests.
• enforce: All links will be rendered to use https:// by default
  • Also redirect port 80 to port 443 for non-default servers in infra_portal

nginx_cert_validity

name: nginx_cert_validity, type: duration, level: G

Nginx self-signed certificate validity, default value is 397d (approximately 13 months).

Modern browsers require website certificate validity to be at most 397 days, hence this default value. Setting a longer validity is not recommended, as browsers may refuse to trust such certificates.

nginx_home

name: nginx_home, type: path, level: G

Nginx server static content directory, default: /www

This is a symlink that actually points to the nginx_data directory. This directory contains static resources and software repository files.

It’s best not to modify this parameter arbitrarily. If modified, it should be consistent with the repo_home parameter.

nginx_data

name: nginx_data, type: path, level: G

Nginx actual data directory, default is /data/nginx.

This is the actual storage location for Nginx static files; nginx_home is a symlink pointing to this directory.

It’s recommended to place this directory on a data disk for easier management of large package files.

nginx_users

name: nginx_users, type: dict, level: G

Nginx Basic Authentication user dictionary, default is an empty dictionary {}.

Format is { username: password } key-value pairs, for example:

nginx_users:
  admin: pigsty
  viewer: readonly

These users can be used to protect certain Nginx endpoints that require authentication.

nginx_port

name: nginx_port, type: port, level: G

Nginx default listening port (serving HTTP), default is port 80. It’s best not to modify this parameter.

When your server’s port 80 is occupied, you can consider using another port, but you need to also modify repo_endpoint and keep node_repo_local_urls consistent with the port used here.

nginx_ssl_port

name: nginx_ssl_port, type: port, level: G

Nginx SSL default listening port, default is 443. It’s best not to modify this parameter.

certbot_sign

name: certbot_sign, type: bool, level: G/A

Use certbot to sign Nginx certificates during installation? Default value is false.

When set to true, Pigsty will use certbot to automatically apply for free SSL certificates from Let’s Encrypt during the execution of infra.yml and deploy.yml playbooks (in the nginx role).

For domains defined in infra_portal, if a certbot parameter is defined, Pigsty will use certbot to apply for a certificate for that domain. The certificate name will be the value of the certbot parameter. If multiple servers/domains specify the same certbot parameter, Pigsty will merge and apply for certificates for these domains, using the certbot parameter value as the certificate name.

Enabling this option requires:

• The current node can be accessed through a public domain name, and DNS resolution is correctly pointed to the current node’s public IP
• The current node can access the Let’s Encrypt API interface

This option is disabled by default. You can manually execute the make cert command after installation, which actually calls the rendered /etc/nginx/sign-cert script to update or apply for certificates using certbot.

certbot_email

name: certbot_email, type: string, level: G/A

Email address for receiving certificate expiration reminder emails, default value is your@email.com.

When certbot_sign is set to true, it’s recommended to provide this parameter. Let’s Encrypt will send reminder emails to this address when certificates are about to expire.

certbot_options

name: certbot_options, type: string, level: G/A

Additional configuration parameters passed to certbot, default value is an empty string.

You can pass additional command-line options to certbot through this parameter, for example --dry-run, which makes certbot perform a preview and test without actually applying for certificates.

DNS

Pigsty enables DNSMASQ on Infra nodes by default to resolve auxiliary names such as i.pigsty, m.pigsty, and api.pigsty; sss.pigsty can optionally provide the Silo endpoint.

Resolution records are stored in the /etc/dnsmasq.d/pigsty/default file on Infra nodes. To use this DNS server, you must add nameserver <ip> to /etc/resolv.conf. The node_dns_servers parameter handles this.

dns_enabled: true                 # setup dnsmasq on this infra node?
dns_port: 53                      # DNS server listen port
dns_records:                      # dynamic DNS records
  - "${admin_ip} i.pigsty"
  - "${admin_ip} m.pigsty supa.pigsty api.pigsty adm.pigsty cli.pigsty ddl.pigsty"

dns_enabled

name: dns_enabled, type: bool, level: G/I

Enable DNSMASQ service on this Infra node? Default value: true.

If you don’t want to use the default DNS server (e.g., you already have an external DNS server, or your provider doesn’t allow you to use a DNS server), you can set this value to false to disable it, and use node_default_etc_hosts and node_etc_hosts static resolution records instead.

dns_port

name: dns_port, type: port, level: G

DNSMASQ default listening port, default is 53. It’s not recommended to modify the default DNS service port.

dns_records

name: dns_records, type: string[], level: G

Dynamic DNS records resolved by dnsmasq, generally used to resolve auxiliary domain names to the admin node. These records are written to the /etc/dnsmasq.d/pigsty/default file on infrastructure nodes.

v4.x default value:

dns_records:
  - "${admin_ip} i.pigsty"
  - "${admin_ip} m.pigsty supa.pigsty api.pigsty adm.pigsty cli.pigsty ddl.pigsty"

The ${admin_ip} placeholder is used here and will be replaced with the actual admin_ip value during deployment.

Common domain name purposes:

• i.pigsty: Pigsty home page
• m.pigsty: Commonly used for the Silo console (optional)
• p.pigsty: Commonly used for the VictoriaMetrics Web UI when explicitly configured in infra_portal
• api.pigsty: API service
• adm.pigsty: Admin service
• Others customized based on actual deployment needs

VICTORIA

Pigsty v4.x uses the VictoriaMetrics suite to replace Prometheus and Loki, providing a superior observability solution:

• VictoriaMetrics: Replaces Prometheus as the time series database for storing monitoring metrics
• VictoriaLogs: Replaces Loki as the log aggregation storage
• VictoriaTraces: Distributed trace storage
• VMAlert: Replaces Prometheus Alerting for alert rule evaluation

vmetrics_enabled: true            # enable VictoriaMetrics?
vmetrics_clean: false             # clean data during init?
vmetrics_port: 8428               # listen port
vmetrics_scrape_interval: 10s     # global scrape interval
vmetrics_scrape_timeout: 8s       # global scrape timeout
vmetrics_options: >-
  -retentionPeriod=15d
  -promscrape.fileSDCheckInterval=5s
vlogs_enabled: true               # enable VictoriaLogs?
vlogs_clean: false                # clean data during init?
vlogs_port: 9428                  # listen port
vlogs_options: >-
  -retentionPeriod=15d
  -retention.maxDiskSpaceUsageBytes=50GiB
  -insert.maxLineSizeBytes=1MB
  -search.maxQueryDuration=120s
vtraces_enabled: true             # enable VictoriaTraces?
vtraces_clean: false              # clean data during init?
vtraces_port: 10428               # listen port
vtraces_options: >-
  -retentionPeriod=15d
  -retention.maxDiskSpaceUsageBytes=50GiB
vmalert_enabled: true             # enable VMAlert?
vmalert_port: 8880                # listen port
vmalert_options: ''               # extra CLI options

vmetrics_enabled

name: vmetrics_enabled, type: bool, level: G/I

Enable VictoriaMetrics on this Infra node? Default value is true.

VictoriaMetrics is the core monitoring component in Pigsty v4.x, replacing Prometheus as the time series database, responsible for:

• Scraping monitoring metrics from various exporters
• Storing time series data
• Providing PromQL-compatible query interface
• Supporting Grafana data sources

vmetrics_clean

name: vmetrics_clean, type: bool, level: G/A

Clean existing VictoriaMetrics data during initialization? Default value is false.

When set to true, existing time series data will be deleted during initialization. Use this option carefully unless you’re sure you want to rebuild monitoring data.

vmetrics_port

name: vmetrics_port, type: port, level: G

VictoriaMetrics listen port, default value is 8428.

This port is used for:

• HTTP API access
• Web UI access
• Prometheus-compatible remote write/read
• Grafana data source connections

vmetrics_scrape_interval

name: vmetrics_scrape_interval, type: interval, level: G

VictoriaMetrics global metrics scrape interval, default value is 10s.

In production environments, 10-30 seconds is a suitable scrape interval. If you need finer monitoring data granularity, you can adjust this parameter, but it will increase storage and CPU overhead.

vmetrics_scrape_timeout

name: vmetrics_scrape_timeout, type: interval, level: G

VictoriaMetrics global scrape timeout, default is 8s.

Setting a scrape timeout can effectively prevent avalanches caused by monitoring system queries. The principle is that this parameter must be less than and close to vmetrics_scrape_interval to ensure each scrape duration doesn’t exceed the scrape interval.

vmetrics_options

name: vmetrics_options, type: arg, level: G

VictoriaMetrics extra command line options, default value:

vmetrics_options: >-
  -retentionPeriod=15d
  -promscrape.fileSDCheckInterval=5s

Common parameter descriptions:

• -retentionPeriod=15d: Data retention period, default 15 days
• -promscrape.fileSDCheckInterval=5s: File service discovery refresh interval

You can add other VictoriaMetrics-supported parameters as needed.

vlogs_enabled

name: vlogs_enabled, type: bool, level: G/I

Enable VictoriaLogs on this Infra node? Default value is true.

VictoriaLogs replaces Loki as the log aggregation storage, responsible for:

• Receiving log data from Vector
• Storing and indexing logs
• Providing log query interface
• Supporting Grafana VictoriaLogs data source

vlogs_clean

name: vlogs_clean, type: bool, level: G/A

Clean existing VictoriaLogs data during initialization? Default value is false.

vlogs_port

name: vlogs_port, type: port, level: G

VictoriaLogs listen port, default value is 9428.

vlogs_options

name: vlogs_options, type: arg, level: G

VictoriaLogs extra command line options, default value:

vlogs_options: >-
  -retentionPeriod=15d
  -retention.maxDiskSpaceUsageBytes=50GiB
  -insert.maxLineSizeBytes=1MB
  -search.maxQueryDuration=120s

Common parameter descriptions:

• -retentionPeriod=15d: Log retention period, default 15 days
• -retention.maxDiskSpaceUsageBytes=50GiB: Maximum disk usage
• -insert.maxLineSizeBytes=1MB: Maximum single log line size
• -search.maxQueryDuration=120s: Maximum query execution time

vtraces_enabled

name: vtraces_enabled, type: bool, level: G/I

Enable VictoriaTraces on this Infra node? Default value is true.

VictoriaTraces is used for distributed trace data storage and query, supporting Jaeger, Zipkin, and other trace protocols.

vtraces_clean

name: vtraces_clean, type: bool, level: G/A

Clean existing VictoriaTraces data during initialization? Default value is false.

vtraces_port

name: vtraces_port, type: port, level: G

VictoriaTraces listen port, default value is 10428.

vtraces_options

name: vtraces_options, type: arg, level: G

VictoriaTraces extra command line options, default value:

vtraces_options: >-
  -retentionPeriod=15d
  -retention.maxDiskSpaceUsageBytes=50GiB

vmalert_enabled

name: vmalert_enabled, type: bool, level: G/I

Enable VMAlert on this Infra node? Default value is true.

VMAlert is responsible for alert rule evaluation, replacing Prometheus Alerting functionality, working with Alertmanager.

vmalert_port

name: vmalert_port, type: port, level: G

VMAlert listen port, default value is 8880.

vmalert_options

name: vmalert_options, type: arg, level: G

VMAlert extra command line options, default value is an empty string.

PROMETHEUS

This section now primarily contains Blackbox Exporter and Alertmanager configuration.

blackbox_enabled: true            # enable blackbox_exporter?
blackbox_port: 9115               # blackbox_exporter listen port
blackbox_options: ''              # extra CLI options
alertmanager_enabled: true        # enable alertmanager?
alertmanager_port: 9059           # alertmanager listen port
alertmanager_options: ''          # extra CLI options
exporter_metrics_path: /metrics   # exporter metrics path

blackbox_enabled

name: blackbox_enabled, type: bool, level: G/I

Enable BlackboxExporter on this Infra node? Default value is true.

BlackboxExporter sends ICMP packets to node IP addresses, VIP addresses, and PostgreSQL VIP addresses to test network connectivity. It can also perform HTTP, TCP, DNS, and other probes.

blackbox_port

name: blackbox_port, type: port, level: G

Blackbox Exporter listen port, default value is 9115.

blackbox_options

name: blackbox_options, type: arg, level: G

BlackboxExporter extra command line options, default value: empty string.

alertmanager_enabled

name: alertmanager_enabled, type: bool, level: G/I

Enable AlertManager on this Infra node? Default value is true.

AlertManager is responsible for receiving alert notifications from VMAlert and performing alert grouping, inhibition, silencing, routing, and other processing.

alertmanager_port

name: alertmanager_port, type: port, level: G

AlertManager listen port, default value is 9059.

If you modify this port, ensure you update the alertmanager entry’s endpoint configuration in infra_portal accordingly (if defined).

alertmanager_options

name: alertmanager_options, type: arg, level: G

AlertManager extra command line options, default value: empty string.

exporter_metrics_path

name: exporter_metrics_path, type: path, level: G

HTTP endpoint path where monitoring exporters expose metrics, default: /metrics. Not recommended to modify this parameter.

This parameter defines the standard path for all exporters to expose monitoring metrics.

GRAFANA

Pigsty uses Grafana as the monitoring system frontend. It can also serve as a data analysis and visualization platform, or for low-code data application development and data application prototyping.

grafana_enabled: true             # enable Grafana?
grafana_port: 3000                # Grafana listen port
grafana_clean: false              # clean data during init?
grafana_admin_username: admin     # admin username
grafana_admin_password: pigsty    # admin password
grafana_auth_proxy: false         # enable auth proxy?
grafana_pgurl: ''                 # external PostgreSQL URL
grafana_view_password: DBUser.Viewer  # PG datasource password

grafana_enabled

name: grafana_enabled, type: bool, level: G/I

Enable Grafana on Infra node? Default value: true, meaning all infrastructure nodes will install and enable Grafana by default.

grafana_port

name: grafana_port, type: port, level: G

Grafana listen port, default value is 3000.

If you need to access Grafana directly (not through Nginx reverse proxy), you can use this port.

grafana_clean

name: grafana_clean, type: bool, level: G/A

Clean Grafana data files during initialization? Default: false.

This operation removes /var/lib/grafana/grafana.db, ensuring a fresh Grafana installation.

If you want to preserve existing Grafana configuration (such as dashboards, users, data sources, etc.), set this parameter to false.

grafana_admin_username

name: grafana_admin_username, type: username, level: G

Grafana admin username, default is admin.

grafana_admin_password

name: grafana_admin_password, type: password, level: G

Grafana admin password, default is pigsty.

IMPORTANT: Be sure to change this password parameter before deploying to production!

grafana_auth_proxy

name: grafana_auth_proxy, type: bool, level: G

Enable Grafana auth proxy? Default is false.

When enabled, Grafana will trust user identity information passed by the reverse proxy (Nginx), enabling single sign-on (SSO) functionality.

This is typically used for integration with external identity authentication systems.

grafana_pgurl

name: grafana_pgurl, type: url, level: G

External PostgreSQL database URL for Grafana persistence storage. Default is an empty string.

If specified, Grafana will use this PostgreSQL database instead of the default SQLite database to store its configuration data.

Format example: postgres://grafana:password@pg-meta:5432/grafana?sslmode=disable

This is useful for scenarios requiring Grafana high availability deployment or data persistence.

grafana_view_password

name: grafana_view_password, type: password, level: G

Read-only user password used by Grafana metadb PG data source, default is DBUser.Viewer.

This password is used for Grafana to connect to the PostgreSQL CMDB data source to query metadata in read-only mode.

3 - Playbook

Pigsty provides three playbooks related to the INFRA module:

• deploy.yml: Deploy the NODE, INFRA, ETCD, MINIO, and PGSQL core modules on all nodes in one pass
• infra.yml: Initialize Pigsty infrastructure on infra nodes
• infra-rm.yml: Remove infrastructure components from infra nodes

deploy.yml

Deploy the NODE, INFRA, ETCD, MINIO, and PGSQL core modules on all nodes in one pass, resolving INFRA/NODE circular dependency issues.

This playbook interleaves subtasks from infra.yml and node.yml, completing deployment of the core components in the following order:

1. id: Generate node and PostgreSQL identities
2. ca: Create self-signed CA on localhost
3. repo: Create local software repository on infra nodes
4. node-init: Initialize nodes and HAProxy
5. infra: Initialize Nginx, DNS, VictoriaMetrics, Grafana, etc.
6. node-monitor: Initialize node-exporter, vector
7. etcd: Initialize etcd (required for PostgreSQL HA)
8. minio: Initialize Silo (optional)
9. pgsql: Initialize PostgreSQL clusters and configure PostgreSQL monitoring

This playbook is equivalent to executing the following five playbooks sequentially:

./infra.yml -l infra    # Deploy infrastructure on infra group
./node.yml              # Initialize all nodes
./etcd.yml              # Initialize etcd cluster
./minio.yml             # Initialize MINIO (Silo) cluster (optional)
./pgsql.yml             # Initialize PostgreSQL clusters

deploy.yml does not currently deploy the Docker module. If Docker is required, set docker_enabled: true and run docker.yml separately.

infra.yml

Initialize the infrastructure module on Infra nodes defined in the infra group of your configuration file.

This playbook performs the following tasks:

• Configures directories and environment variables on Infra nodes
• Downloads and creates a local software repository to accelerate subsequent installations
• Incorporates the current Infra node as a common node managed by Pigsty
• Deploys infrastructure components (VictoriaMetrics/Logs/Traces, VMAlert, Grafana, Alertmanager, Blackbox Exporter, etc.)

Playbook notes:

• This is an idempotent playbook - repeated execution will overwrite infrastructure components on Infra nodes
• To preserve historical monitoring data, set vmetrics_clean, vlogs_clean, vtraces_clean to false beforehand
• Unless grafana_clean is set to false, Grafana dashboards and configuration changes will be lost
• When /www/pigsty/repo_complete exists, this playbook skips internet downloads; the file is the SHA-256 manifest and completion marker generated by SOW
• Complete execution takes approximately 1-3 minutes, depending on machine configuration and network conditions

Available Tasks

# ca: create self-signed CA on localhost files/pki
#   - ca_dir        : create CA directory
#   - ca_private    : generate ca private key: files/pki/ca/ca.key
#   - ca_cert       : signing ca cert: files/pki/ca/ca.crt
#
# id: generate node identity
#
# repo: bootstrap a local yum repo from internet or offline packages
#   - repo_dir      : create repo directory
#   - repo_check    : check repo exists
#   - repo_prepare  : use existing repo if exists
#   - repo_build    : build repo from upstream if not exists
#     - repo_upstream    : handle upstream repo files in /etc/yum.repos.d
#       - repo_remove    : remove existing repo file if repo_remove == true
#       - repo_add       : add upstream repo files to /etc/yum.repos.d
#     - repo_url_pkg     : download packages from internet defined by repo_url_packages
#     - repo_cache       : make upstream yum cache with yum makecache
#     - repo_boot_pkg    : install sow and dnf/yum download utilities
#     - repo_pkg         : download packages & dependencies from upstream repo
#     - repo_create      : atomically create RPM/APT metadata with sow create --pigsty
#     - repo_use         : add newly built repo into /etc/yum.repos.d
#   - repo_nginx    : launch a nginx for repo if no nginx is serving
#
# node/haproxy/monitor: setup infra node as a common node
#   - node_name, node_hosts, node_resolv, node_firewall, node_ca, node_repo, node_pkg
#   - node_feature, node_kernel, node_tune, node_sysctl, node_profile, node_ulimit
#   - node_data, node_admin, node_timezone, node_ntp, node_crontab, node_vip
#   - haproxy_install, haproxy_config, haproxy_launch, haproxy_reload
#   - haproxy_register, node_exporter, node_register, vector
#
# infra: setup infra components
#   - infra_user     : setup infra os user group
#   - infra_dir      : create infra data/config/runtime directories
#   - infra_env      : env_patroni, env_pg, env_pgadmin, env_etcd, env_pglog, env_var
#   - infra_pkg      : install infra packages
#   - infra_cert     : issue cert for infra components
#   - dns            : dns_config, dns_record, dns_launch
#   - nginx          : nginx_dir, nginx_config, nginx_cert, nginx_static, nginx_launch, nginx_certbot, nginx_reload, nginx_exporter
#   - victoria       : vmetrics/vlogs/vtraces clean, config & launch; vmalert_config, vmalert_launch
#   - alertmanager   : alertmanager_config, alertmanager_launch
#   - blackbox       : blackbox_config, blackbox_launch
#   - grafana        : grafana_clean, grafana_dir, grafana_config, grafana_launch, grafana_provision
#   - infra_register : add_metrics, add_logs, add_ds

infra-rm.yml

Remove Pigsty infrastructure from Infra nodes defined in the infra group of your configuration file.

Common subtasks include:

./infra-rm.yml               # Run all phases: deregister, stop, remove config/environment/data, and uninstall packages
./infra-rm.yml -t deregister # Only deregister monitoring targets, Grafana datasources, and Nginx log collection
./infra-rm.yml -t service    # Stop infrastructure services on INFRA
./infra-rm.yml -t config     # Remove INFRA configuration and systemd units
./infra-rm.yml -t env        # Remove admin environment files
./infra-rm.yml -t data       # Remove INFRA data
./infra-rm.yml -t package    # Uninstall INFRA packages

4 - Monitoring

This document describes monitoring dashboards and alert rules for the INFRA module in Pigsty.

Dashboards

Pigsty provides the following monitoring dashboards for the Infra module:

Alert Rules

Pigsty provides the following two alert rules for the INFRA module:

You can modify or add new infrastructure alert rules in files/victoria/rules/infra.yml.

Alert Rule Configuration

################################################################
#                Infrastructure Alert Rules                    #
################################################################
- name: infra-alert
  rules:

    #==============================================================#
    #                       Infra Aliveness                        #
    #==============================================================#
    # infra components (victoria,grafana) down for 1m triggers a P1 alert
    - alert: InfraDown
      expr: infra_up < 1
      for: 1m
      labels: { level: 0, severity: CRIT, category: infra }
      annotations:
        summary: "CRIT InfraDown {{ $labels.type }}@{{ $labels.instance }}"
        description: |
          infra_up[type={{ $labels.type }}, instance={{ $labels.instance }}] = {{ $value  | printf "%.2f" }} < 1

    #==============================================================#
    #                       Agent Aliveness                        #
    #==============================================================#

    # agent aliveness are determined directly by exporter aliveness
    # including: node_exporter, pg_exporter, pgbouncer_exporter, haproxy_exporter
    - alert: AgentDown
      expr: agent_up < 1
      for: 1m
      labels: { level: 0, severity: CRIT, category: infra }
      annotations:
        summary: 'CRIT AgentDown {{ $labels.ins }}@{{ $labels.instance }}'
        description: |
          agent_up[ins={{ $labels.ins }}, instance={{ $labels.instance }}] = {{ $value  | printf "%.2f" }} < 1

5 - Metrics

Note: Pigsty v4.0 has replaced Prometheus/Loki with VictoriaMetrics/Logs/Traces. The following metric list is still based on v3.x generation, for reference when troubleshooting older versions only. To get the latest metrics, query directly in https://p.pigsty (VMUI) or Grafana. Future versions will regenerate metric reference sheets consistent with the Victoria suite.

INFRA Metrics

The INFRA module has 964 available metrics.