Skip to content

Document installer-backed node maintenance and recovery#135

Open
HDauven wants to merge 4 commits into
mainfrom
agent/operator-reliability-runbook
Open

Document installer-backed node maintenance and recovery#135
HDauven wants to merge 4 commits into
mainfrom
agent/operator-reliability-runbook

Conversation

@HDauven

@HDauven HDauven commented Jul 14, 2026

Copy link
Copy Markdown
Member

Summary

  • replace the generic maintenance page with an actionable health, resource, log, update, reboot, key-management, and incident-response runbook
  • align upgrade and rollback commands with node-installer v0.5.22, including role/network flags, staged downloads, managed-file replacement, target-user behavior, and explicit restart verification
  • make fast-sync and recovery guidance role-aware and stop recommending routine unstaking, blind ruskreset, or source-built wallet updates
  • update archive installation and canonical GraphQL checks, with an explicit boundary between execution-state snapshots and complete historical archive indexes
  • correct consensus-key installation, explorer references, and the transaction-lifecycle description of installer-configured mempool expiry

Closes #21.

Why

The existing maintenance page did not provide the checks or recovery actions requested by the issue. Adjacent pages had also diverged from the current installer:

  • stopping Rusk before running the installer defeated its staged-download update flow;
  • archive operators could accidentally install the default binary by omitting --feature archive;
  • rusk.toml, the systemd unit, helper scripts, and wallet configuration were described as though updates preserved local changes;
  • download_state was presented as an archive bootstrap even though it restores execution state rather than pre-snapshot historical indexes; and
  • several procedures omitted required root privileges or recommended destructive reset/source-build paths too early.

Validation

  • traced node-installer main at e25a191 and released v0.5.22 at ba73650, including install staging, binary verification, file ownership, preserved files, systemd behavior, log rotation, download_state, and ruskreset
  • ran bash -n/sh -n over the released installer and helper scripts
  • ran the installer-managed ruskquery helper against live mainnet for info, peers, block-height, and installer version; live Rusk reported 1.7.1
  • exercised canonical archive GraphQL introspection and checkBlock(... onlyFinalized: true) against the live archive endpoint
  • traced Rusk state export and archive storage separately to verify that a state snapshot does not backfill archive indexes
  • confirmed Rusk built-in mempool defaults against source and node-installer mainnet/testnet overrides against v0.5.22
  • npm run build (Astro build and all internal links)
  • Chromium full-page inspection at 1440x1000 and 390x844 for maintenance, upgrade, archive, and troubleshooting pages

@HDauven HDauven requested a review from Neotamandua July 15, 2026 00:34
@HDauven HDauven marked this pull request as ready for review July 15, 2026 00:34
@HDauven HDauven requested a review from GeorgianDusk July 15, 2026 07:56
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Content: Maintenance & monitoring of a server running a Dusk node

1 participant