taler-monitoring/site-gen

monitoring site-gen

Build console-style HTML from taler-monitoring runs and stage them for host Caddy (/monitoring, /monitoring_err).

Same global reporting defaults as host-agents (run-host-report.sh):

• RUN_TIMEOUT=600 on each suite run
• line-buffered logs (continuous write)
• always write HTML (first run / empty logs too)
• Forgejo commit link in the page header

Applies to all nine fronts (GOA + FP stage + FP prod) in the default SITES list.

Host	Role
firecuda-external	optional runner (public DNS)
koopa-external	deploy staging + Caddy (root service)
host-agents	primary: GOA + FP on-box (../host-agent/)

No secrets in settings.conf — only hosts, paths, site list.

firecuda-external (outside-only) — disabled for now

firecuda-external could run public-only monitoring (SKIP_SSH=1, phases urls versions) on a schedule (launchd every 4h) and write HTML under ~/var/taler-monitoring-sites-work/html/. Scripts for that still live here (install-firecuda-timer.sh, run-on-firecuda.sh, plist template).

Current policy: do not run a real launchd job on firecuda.
Primary GOA monitoring is koopa host-agent (host-agent/, user hernani, local podman exec). Re-enable firecuda only if you explicitly want a second, outside-only view.

# optional one-shot (no timer):
# ssh firecuda-external '~/taler-monitoring-site-gen/site-gen/run-on-firecuda.sh'

# if a timer was installed earlier, ensure it is off:
# ssh firecuda-external 'launchctl list | grep taler-monitoring || echo off'

Disabled plists may remain as *.plist.disabled-* under ~/Library/LaunchAgents/.

Quick start (laptop)

cd site-gen
cp settings.conf.example settings.conf   # optional
chmod +x generate-monitoring-sites.sh deploy-monitoring-sites.sh console_to_html.py

ONLY_HOSTS="bank.hacktivism.ch exchange.hacktivism.ch taler.hacktivism.ch" \
  SKIP_SSH=1 RUNNER_SSH= \
  ./generate-monitoring-sites.sh

./deploy-monitoring-sites.sh

Then on koopa as root: see ROOT-ON-KOOPA.md.

URL rules

Path	When
/monitoring_err/	always: full console log + error list on top (CONTINUE_ON_ERROR run)
/monitoring/	clean full log if last strict run exit 0; else stub linking to /monitoring_err/

Footer links Forgejo commit of the admin-log tree used for that run.

Layout

site-gen/
  settings.conf.example
  generate-monitoring-sites.sh
  deploy-monitoring-sites.sh
  console_to_html.py
  caddy-monitoring-handles.snippet
  ROOT-ON-KOOPA.md
  README.md