counterweight 4bee182978 ntfy-emergency-app, vaultwarden, forgejo: use the caddy_site role ... The plain reverse_proxy shape. All three removed a byte-identical 23-line block (verified by md5 of the diff with the service name normalised) and gained the same 7-line include_role call. The caddy_sites_dir self-reference goes with it. Verified in check mode, nothing applied to the hosts yet: - ntfy-emergency-app: site task ok, changed=0 - vaultwarden: site task ok; the one changed task is a pre-existing always-restarts fail2ban step, identical before the edit - forgejo: check mode cannot run this playbook at all - get_url does not download in check mode so the next task fails on "Source /tmp/forgejo not found". Confirmed identical before the edit. Covered instead by the Stage 2 dry-run, which ran the role against vipy with forgejo's real parameters and reported ok/unchanged. All 14 site files on all 3 hosts still byte-identical. Real runs for these three are still outstanding. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>		2026-09-11 23:24:57 +02:00
..
deploy_ntfy_emergency_app_playbook.yml	ntfy-emergency-app, vaultwarden, forgejo: use the caddy_site role	2026-09-11 23:24:57 +02:00
ntfy_emergency_app_vars.yml	vars: derive remote_host_name from role groups instead of hostnames	2026-09-11 22:00:43 +02:00