tos.md was baked into the repo and read once at startup, so changing
the terms required a code change and deploy. It's now a Redis-backed
singleton (models/tos.js), editable from a new "Terms of Service" card
on the admin Dashboard, with the bundled tos.md used only as a
one-time seed for new deployments.
- routes/tos.js: GET (any authenticated user) / PUT (app_sso_admin
only) via /api/tos. Saving can optionally reset every user's
tos_accepted flag so they're asked to re-accept -- off by default,
since a wording fix shouldn't re-prompt everyone.
- routes/index.js: /tos and /onboarding now render the live content
instead of a module-level constant computed once at process start.
ops/backup.sh snapshots LDAP (slapcat), Redis (BGSAVE, dynamic RDB path
lookup), and ./config for standalone deployments, with retention. A
background service polls GitHub releases every 24h and surfaces an
admin-only banner in the UI when a newer version is published.
The Integrations page's Service Accounts (bind-only, organizationalRole)
don't cover the other real use case: an account something actually runs
as on a Linux host -- a media manager, a torrent client, Emby -- with a
real uidNumber/gidNumber that owns files, and a group other accounts
join for write access (e.g. a `stuff_manager` group granting write
rights to a media library). That needs a real posixAccount, which the
bind-only model can't be.
- New well-known group `app_sso_service_account`, seeded the same way as
app_sso_admin/app_sso_invite/app_sso_oauth_admin (docker-entrypoint.sh,
ops/ldap-setup.sh). Not a permission gate -- a marker.
- "Add new user" form gets a "This is a service account" checkbox: swaps
the person-shaped fields (first/last name, birthday, ToS agreement)
for a single account-name field, since none of those make sense for a
non-person account. On create, the route adds the user to
app_sso_service_account.
- User.listDetail() annotates each user with isServiceAccount (checked
against the marker group's member list once per call, not the memberof
overlay's reverse attribute -- not reliably returned by every LDAP
server this app might point at, confirmed against a real external
directory during testing). Users page shows a "service" badge.
- Notification broadcasts (filter_type=all/all_active) exclude service
accounts by default -- nobody reads mail as `stuff_manager`.
- Fixed a real, previously-unrelated bug this surfaced: addPosixAccount
unconditionally set `mail: data.mail` in the LDAP entry even when
undefined, and ldapts/slapd reject an attribute given an explicit
undefined value ("no values for attribute type") rather than treating
it as absent. This meant creating ANY user without an email already
failed outright -- not something a service account (which commonly has
no real mailbox) could route around. Made mail conditional, matching
how mobile/sshPublicKey/dob already work.
- docs/ldap.md now explains both kinds of service account side by side
and when to use which.
Verified end-to-end against a real external LDAP server (not a local
sandbox): created a service account with no email, confirmed it's
correctly flagged and excluded from broadcast recipient resolution,
confirmed a normal user is unaffected, confirmed the code degrades
gracefully if the marker group doesn't exist yet (pre-upgrade
deployments).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- OAuth Apps and LDAP Info are both "how do other apps/hosts plug into
this SSO" concerns -- merged into a single /integrations page with
tabs, replacing the two separate nav items with one. /oauth-clients
and /ldap-info 301-redirect there for compat.
- Add a Service Accounts section under the LDAP tab: bind-only LDAP
identities (organizationalRole + simpleSecurityObject, no
posixAccount) for apps/hosts, as opposed to real people. Create,
rotate password, and delete, all from the UI -- previously the only
such account (theta-env's bootstrap-created cn=ldapclient) was
invisible to the Users page entirely (filtered out by
conf.ldap.userFilter) and had no GUI way to see or rotate it; the new
ServiceAccount model uses the exact same objectClasses bootstrap.js
already creates cn=ldapclient with, so it recognizes and manages that
account too, not just ones created through this UI.
- The ldap-client bash snippet now points at "create one under Service
Accounts above" instead of a bare textual example.
Verified against a real LDAP server (not just the dev sandbox's usual
unreachable one): created a service account, confirmed it binds
successfully with the generated password, rotated its password, and
deleted it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
New admin-only page (nav: "LDAP Info") that answers "what do I put in my
app's LDAP settings" without reading a doc: LDAPS URL, base DN, user/group
search bases, user filter, username attribute, and an example bind DN, all
derived from the running conf.ldap + request host rather than hardcoded --
so it's always correct for the actual deployment, copy-button on every
field.
Also generates a copy-pasteable bash snippet that clones
theta42/ldap-client and writes its ldap.vars file with the real host/base
DN/sso_url already filled in (bind password and SSO API token left as
placeholders with inline instructions, since those need to be created,
not derived).
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The Compose form's "Send to" radio group had "All active users" checked
by default with no confirmation before Send -- anyone opening the
Dashboard to see how the feature works, typing a test subject/message,
and clicking Send would broadcast to every active user. Remove the
default (a target must now be explicitly chosen) and require a confirm
step before actually sending to "all" or "all_active".
Also add a hard safety net in models/email.js: Mail.send is a no-op
under NODE_ENV=test, so the automated test suite (which exercises the
real notification/password-reset/invite/OTP-by-email routes with
NODE_ENV=test) can never deliver real mail regardless of what recipient
list a test resolves.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
It linked to fixed personal infrastructure (718it.biz, vm42.us,
git.theta42.com) with an admin's name in one label — not appropriate for
what should be a generic, publicly-releasable template. Flagged in
theta42/sso-manager-node#45, tracked as #46; resolves that issue.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
The admin UI is a traditional multi-page app that loads ~13 separate
vendor/app JS+CSS files on every full navigation; none were compressed and
Cache-Control was max-age=0 (Express's default), forcing a revalidation
round-trip for every asset on every page view. Add gzip (compression
middleware) and sane Cache-Control (7d for vendor libs under
/static-modules, 1h for the app's own /static JS/CSS, which isn't
cache-busted). Matches the equivalent fix in theta42/proxy.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
theta42/proxy fronts an arbitrary number of hosts behind SSO, each with its
own callback URL (https://<host>/__proxy_auth/callback) — proxy's own code
comment already assumed "a wildcard redirect URI covers all", but no
wildcard matching existed here, so every proxied host's callback had to be
registered on the shared OAuth client individually or /oauth/authorize
would reject it with InvalidRedirectURI.
Add `*` (one hostname label) / `**` (any number of labels) wildcard support
to redirect_uri matching, e.g. `https://**.example.com/__proxy_auth/callback`
now covers every host proxy fronts under example.com. Exact matches still
work exactly as before.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Replace the separate Profile/API Tokens nav items with a single link
showing the logged-in user's name, pointing at their own profile.
- Merge admin.ejs + notifications.ejs into a new dashboard.ejs page.
/admin and /notifications now 301-redirect to /dashboard.
- Fold the API Tokens page into profile.ejs as a self-service-only
section, gated on isOwnProfile so it never appears when an admin
views another user's profile via /users/:uid. /api-tokens 301s to /.
- Fix: the section must not carry class="row" — app-base.js runs a
page-wide $('div.row').fadeIn() on every page load that would reveal
it regardless of the isOwnProfile check, since it fires before this
page's own gating logic.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Reported: creating any user via the API failed with
{"name":"InvalidSyntaxError","message":"gidNumber: value #0 invalid per syntax Code: 0x15"}
Root cause: addPosixGroup() computes the next gidNumber as
`Math.max(...groups.map(i => i.gidNumber)) + 1`. theta-env's
bootstrap.js creates the first admin via raw ldapadd with a hardcoded
uidNumber/gidNumber (10000) directly on the user entry, but never
creates a matching posixGroup entry -- so on a theta-env-bootstrapped
directory there are zero posixGroup entries, `Math.max()` on an empty
array is `-Infinity` in JS (not 0), and `-Infinity + 1` stringifies to
"-Infinity" -- an invalid LDAP integer, rejected by the directory. This
broke every single user creation, not just this one.
Separately: the reporter's intended scheme is for organically-created
users to start at uidNumber/gidNumber 1500, distinct from the
bootstrap admin's reserved 10000. Fixing the crash with a bare "floor
of 1500" alone wouldn't achieve that, since addPosixAccount's own
Math.max() would still find the admin's posixAccount entry (uidNumber
10000, found via a different, correctly-indexed search) and allocate
10001 for the next user.
Added a shared nextPosixId(entries, key) helper: takes the highest
existing value strictly below conf.ldap.uidGidReservedFloor (default
9000) plus one, or conf.ldap.uidGidMin (default 1500) if there are no
such entries. Ids at/above the reserved floor -- like the bootstrap
admin's 10000 -- are ignored entirely when computing the next
available number, so real users always start at 1500 and grow upward
regardless of the admin's reserved id.
Verified against a real theta-env deployment end to end:
- Reproduced the exact reported crash on a fresh bootstrap
- After the fix: first real user gets uidNumber/gidNumber "1500",
second gets "1501" -- admin's 10000 never enters the calculation
- New unit tests (nodejs/tests/posix_id.test.js, no LDAP required):
6/6 pass, covering the empty-array case, the reserved-floor
exclusion, and the NaN-from-missing-value case
- npm test: 18/18 passing tests still pass (unchanged); the other 155
failures are pre-existing/environmental (no LDAP server in this
sandbox) -- confirmed via git stash before starting this fix
* Fix commit hash not showing in Docker builds
build_info.js computed buildHash via `git rev-parse --short HEAD` at
runtime, but the final image intentionally has no git binary and no
.git directory (kept lean, per .dockerignore) — so this always failed
silently and the footer's version line showed "unknown" for every
Docker deployment. Working correctly only for bare-metal/dev, where
git + .git are actually present.
Added a throwaway gitinfo build stage that reuses the main base image
(no extra pull) with git installed just for this stage, reads .git
from the build context (now no longer excluded — see .dockerignore),
and bakes the resolved short hash into a small file that IS copied
into the final image. build_info.js reads that file first, falling
back to the old git-rev-parse behavior (still needed for bare-metal).
Verified against a real build: `docker exec sso-manager cat
/app/.build_commit` matches `git rev-parse --short HEAD` on the host,
and the footer now shows the real hash instead of "unknown". Same fix
already applied to proxy (theta42/proxy#133).
* Support GIT_COMMIT build-arg override for submodule builds
The gitinfo stage from the previous commit works for a standalone
clone (.git is a real directory) but not when this repo is built as a
git submodule (e.g. from theta-env): a submodule's .git is a pointer
FILE, not a directory — the real object database lives in the
superproject's .git/modules/, outside this repo's own directory and
therefore outside Docker's build context entirely. `git rev-parse`
can never resolve it from in here no matter what, so builds via
theta-env still baked in "unknown" despite the earlier fix.
Add an optional GIT_COMMIT build-arg that, when set, wins over the
in-context git resolution. theta-env's setup.sh now computes it on the
host (where the submodule DOES resolve correctly) and passes it via
docker-compose.yml's build.args. Same fix in proxy: theta42/proxy#133.
Verified via theta-env's actual setup.sh end to end: rebuilding with
this change, `docker exec sso-manager cat /app/.build_commit` now
matches `git -C sso-manager-node rev-parse --short HEAD` on the host
(previously: "unknown").
Auditing mobile/responsive views turned up two classes of bug:
1. Missing .table-responsive wrapper (token.ejs and profile.ejs's other
table already had it): admin.ejs (2 tables), invites.ejs (6 columns),
notifications.ejs (5 columns), users.ejs (7 columns — the worst case).
Without it these overflow the page horizontally on narrow viewports
instead of scrolling within the table.
2. groups.ejs's search/sort/count toolbar was a plain `d-flex` (no
flex-wrap) containing a search input, a <select> with a hard
min-width:175px, and a nowrap count span — on a narrow viewport
there's nowhere for that content to go but off-screen. Added
flex-wrap and gave the search input a flex-basis so it wraps onto
its own line first when space is tight.
Verified: EJS compiles for all five templates, npm test 192/192 pass,
and fetched each route from a running instance to confirm the fixes
are present in the served HTML.
* docs: cleanup for public release (fix stale/wrong API docs, LICENSE, versions)
Documentation cleanup ahead of the public release announcement. Fixes a set
of confirmed issues from a prior audit:
- LICENSE: fill in MIT template placeholders (theta42, 2026).
- README.md: fix broken API docs link (api.md -> API.md), correct required
Node.js version (13.x -> 20.x), add the missing app_sso_invite group to
the LDAP groups table, scrub hardcoded dc=theta42,dc=com to the generic
dc=example,dc=com used elsewhere, add a "Recommended: Docker or
install.sh" section pointing to DEPLOYMENT.md/docs before the manual
OpenLDAP walkthrough, and drop an emoji from a warning callout.
- nodejs/api.md: deleted — it was a stale/legacy doc with wrong routes,
wrong request bodies, and endpoints that are dead/commented-out code.
The root API.md is the accurate, current reference; README now links
there directly.
- API.md: add the missing app_sso_invite permission group, fix the
documented invite response to match the real {token, link, mail_sent}
payload, document the previously-undocumented GET/PUT/DELETE
/api/user/invite endpoints, add the real allowed_groups field to the
OAuth client management examples, and document POST /api/oauth/authorize
(the endpoint that actually issues the code after consent).
- nodejs/routes/auth.js + API.md: fix "emaill address" typo in the
password-reset response message (source and docs kept in sync).
- DEPLOYMENT.md: fix the top-level summary to mention Redis, matching
docs/deployment.md and the entrypoint behavior it already documents.
Flagged, not changed: tos.md reads like a personal home-lab acceptable-use
policy (Emby/Gitea/Proxmox/Discord/Signal, first-person "the admin") rather
than generic OSS docs. Left in place pending a manual decision to
genericize, relocate, or remove it.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
* docs: genericize tos.md template, track runtime-editable terms in #39
Removes operator-specific references (Emby, Gitea, Proxmox, Discord,
Signal, first-person "the admin") so the shipped tos.md reads as a
neutral starting template rather than one operator's internal policy.
Actual runtime editability (admin/legal editing terms without a code
change) is tracked in issue #39, not implemented here.
---------
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
- api_tokens.ejs: created_on/last_used_on come back from Redis as strings
(model-redis only coerces fields with an explicit `type`), so `new Date(ms)`
yielded "Invalid date". Use `moment(ms, "x")` (the hosts.ejs/dns.ejs
precedent) which parses a numeric string-or-number as a Unix-ms timestamp.
- api_tokens.ejs: `isExpired` is a class getter not serialized to the client
JSON, so the "expired" badge never showed — compute expiry in the view via
`Date.now() > Number(expires_at)`. Also guard the `last_used_on: 0` / falsy
case (string "0" is truthy) so unset timestamps render "—" not "1970".
- middleware/auth.js: authIO did `checkToken(socket.handshake.auth.token || 0)`,
so any socket connect without a token (login page, pre-login) did an
`AuthToken.get(0)` lookup and logged a noisy `EntryNotFound` trace. Guard:
reject the socket with a generic 401 when there's no token (behavior-
preserving — unauth sockets were already rejected; just no Redis lookup / 404).
Co-authored-by: Claude <noreply@anthropic.com>
Personal access tokens so scripts/CI can call the management API without a
browser session. Each logged-in user mints their own token; it authenticates as
the creator (carries their LDAP group permissions, re-resolved live), so the
existing permission.byGroup checks apply unchanged.
- models/api_token.js: new ApiToken model (sso_<id>_<secret> format; id is the
lookup key, secret bcrypt-hashed + isPrivate, shown once). add()/rotate()/
authenticate(); optional expires_at; best-effort last_used_on. No _ttl
(persists; lifetime via expires_at).
- routes/api_token.js: self-service CRUD (list/get/update/delete/rotate),
owner-scoped (created_by === req.user.uid, 403 otherwise).
- middleware/auth.js + models/auth.js: accept `Authorization: Bearer sso_...`
(precedence over the auth-token session header); checkApiToken collapses
every failure to one generic 401 (no existence/secret/expiry leak).
- views/api_tokens.ejs + routes/index.js (GET /api-tokens): self-service page
(forceLogin, no group gate) — create (token shown once), edit, rotate, revoke.
- views/top.ejs: "API Tokens" nav entry visible to all logged-in users.
- public/js/app.js: app.apiToken client module.
- DEPLOYMENT.md + docs/deployment.md: API tokens section.
Co-authored-by: Claude <noreply@anthropic.com>
wmantly
