Upgrades
This guide is auto-synced from
docs/upgrades.mdin the pocket-automations repo. Edit it there — changes here are overwritten on every build.
Use playbooks/upgrade.yml for day-2 component updates after the initial deployment.
The upgrade playbook is intentionally scoped to deployed software and configuration. It does not perform backup, restore, disaster recovery, staking, unstaking, or supplier lifecycle operations.
Full Upgrade
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.ymlThe full run updates components in this order:
- common host prerequisites and Docker runtime checks;
pocketdCLI;- Igniter Provider;
- optional Igniter Middleman;
- HA RelayMiner relayers and miners;
- Prometheus/Grafana monitoring;
- Caddy reverse proxy;
- validation report.
Component Upgrades
Run a single component with tags:
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.yml --tags pocketd
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.yml --tags igniter
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.yml --tags relayminer
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.yml --tags monitoring
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/upgrade.yml --tags reverse_proxyValidation can be run separately:
ansible-playbook -i inventories/generated/<host-name>/hosts.yml playbooks/validate.ymlVersion Pins
Prefer explicit version pins for production changes:
pocketd_version: vX.Y.Z
igniter_repo_version: <tag-or-commit>
ha_relayminer_repo_version: <tag-or-commit>
monitoring_prometheus_image: prom/prometheus:<tag>
monitoring_grafana_image: grafana/grafana:<tag>Using main or latest is convenient for labs but should be treated as an intentional production decision.
HA RelayMiner Rolling Strategy
playbooks/upgrade.yml uses serial execution for HA RelayMiner by default:
upgrade_relayminer_serial: 1This updates one relayer/miner host at a time. Increase only if the topology can tolerate simultaneous component restarts.
Post-Upgrade Checks
After upgrades, confirm:
playbooks/validate.ymlreports noFAILitems;- relayer health responds at
/health; - relayer and miner metrics respond;
- Prometheus and Grafana are healthy;
- public relay TLS still works;
- Grafana public endpoint, if enabled, still requires authentication;
- Igniter Provider UI/workflows are healthy before running any lifecycle operation.