• v1.24.0 eb10d684d8

    LDAP WebGUI v1.24.0
    All checks were successful
    Tests / pytest (Python 3.11) (push) Successful in 2m36s
    Tests / pytest (Python 3.12) (push) Successful in 2m41s
    Tests / shellcheck (install.sh, update.sh, bootstrap-ldap.sh, webupdate.sh, get_ldap-webgui.sh, setup.sh) (push) Successful in 2m41s
    Release / Publish release (push) Successful in 3s
    Stable

    Lyra released this 2026-11-01 19:50:27 +01:00 | 22 commits to main since this release

    Added

    • Who may reach the WebGUI. nginx now admits loopback plus the hosts
      and networks in /etc/nginx/ldap-webgui-access.conf, and nothing else;
      /health additionally admits the monitoring hosts in
      /etc/nginx/ldap-webgui-monitoring.conf. The guided setup asks for both
      (Access, suggesting the address of your SSH session); install.sh
      takes --allow LIST and --allow-monitoring LIST (IPv4/IPv6, CIDR —
      a network with host bits set is refused with the corrected form).
      any writes an unrestricted list and needs a second, explicit
      confirmation (typed yes, access_any_keep=yes, or --confirm-any).
      Loopback is always in the list — the health checks of the setup and of
      every update go through https://localhost/health.
    • Settings → Access shows both lists read-only, whether the host is
      restricted, unrestricted or in an unknown state, and which entry admits
      your own address.
    • README → Operating recommendation: admin network or VPN only, never
      internet-facing; ideally from a jump host, at least from a browser
      profile that is not the everyday one.

    Changed

    • install.sh stops when nginx -t fails after it wrote an access
      file
      , and puts the previous file back — before, a failing nginx -t
      was only a warning. A re-install that keeps both files still warns.
    • install.sh without --allow keeps an existing access file; where
      there is none it writes loopback only and says how to open it.
    • The vhost no longer has an ACME /.well-known/acme-challenge/
      exception on port 80, and no longer recommends certbot — certificates
      come from your own CA.
    • Port 80's redirect and /robots.txt answer only allowed hosts. A
      plain return runs before nginx's allow/deny; both now go through
      try_files to a named location, which is reached only after the
      access check. Carry this over with the include lines.

    Fixed

    • An existing directory's CA certificate is trusted by the system.
      setup.sh (directory existing) checked the CA as root, then wrote its
      path into config.yaml unchanged — a CA under /root passed the check,
      but the service (ldapwebgui, ProtectHome=true) could not read it,
      /health answered 503 and the setup stopped at start. Unless the
      system already trusts it (same SHA-256 fingerprint in
      /etc/ssl/certs/ca-certificates.crt, which ldap.tls.ca_cert then
      names), the setup copies it to
      /usr/local/share/ca-certificates/ldap-webgui-ca.crt, runs
      update-ca-certificates and names that file. It stays trusted if the
      WebGUI is removed: delete the file and run
      update-ca-certificates --fresh to undo.
    • setup.sh says why the WebGUI is not healthy. When /health does
      not answer 200 with the new version, the error now shows the last
      answer — the HTTP status and the start of the body, e.g. HTTP 503: {"ldap": "unreachable", ...} — instead of only pointing at the journal
      and app.log, which do not record a failed LDAP connection.

    Upgrade notes

    • nginx/ldap-webgui.conf changed. update.sh never installs the
      vhost, so an updated host stays unrestricted until you change it by
      hand; the update's nginx drift check says so. Order matters — create
      the files first, or nginx -t fails on the missing includes:

      printf 'allow 127.0.0.1;\nallow ::1;\nallow 192.168.10.0/24;\n' \
          > /etc/nginx/ldap-webgui-access.conf        # your admin hosts/networks
      printf '# monitoring hosts, /health only\n' \
          > /etc/nginx/ldap-webgui-monitoring.conf
      

      Then add include /etc/nginx/ldap-webgui-access.conf; and deny all;
      to both server blocks, and to location = /health the two includes
      (access, then monitoring) followed by deny all; — exactly as in the
      shipped nginx/ldap-webgui.conf. nginx -t && systemctl reload nginx.
      Keep 127.0.0.1 and ::1: without them every update rolls back.

    • Re-running install.sh on such a host copies the new vhost and,
      without --allow, writes loopback only — pass --allow with your admin
      hosts.

    Update without git access: download the source tar.gz below, extract it, then run the update.sh it contains:
    sudo bash <extracted-dir>/update.sh --archive <tarball>

    Downloads