An instance's isolation never needed a domain — its own slug, dir, secrets,
IP and randomly-allocated host port already make two copies independent. But
the routing layer assumed one, so a LAN-only box got a broken instance rather
than a port-served one. Four fixes:
- instanceCreate now rewrites the parent-service column of the cloned config's
PORT_ rows to match the service names it stamps into the compose. That value
is stored as network_resources.parent_service and joined against the
compose-derived service names, so an instance left carrying the TYPE's
service name matched nothing: it rendered in the WebUI with no port, no URL
and no login row despite being up and reachable.
- `instance create --local` (plus a LAN-only toggle in the modal) forces every
port to access=private, traefik=false, for a second copy that should stay
off the domain even when one is configured.
- initializeAppVariables forces the traefik column false when no CFG_DOMAIN_n
is set. Previously a traefik=true port with an empty domain stamped
Host(`app.`) — a trailing-dot host matching nothing — and dragged APP_URL to
https://app. with it, breaking every app that builds its links from APP_URL.
host_setup is blanked for the same reason. The published host port is
untouched; access type, not the traefik flag, gates allocation.
- APP_URL's direct host-port branch now prefers a new $local_ip_v4 (the source
IP for the default route) over $public_ip_v4, which is the WAN address from
an external resolver. LibrePortal never forwards ports, so the WAN address
was unreachable for exactly the LAN/VPN clients that branch serves.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Instance install (bugs found by running one end to end):
- The cloned compose kept the TYPE's tag namespace
(#LIBREPORTAL|BOOKSTACK_APP_KEY_1_TAG|...) while the config had been
re-namespaced to CFG_<SLUG>_*, so tagsProcessorAppConfigValues matched
nothing, the placeholders survived and the pre-start guard refused to
launch. Rewrite the tag names and *_DATA tokens too — narrowly, so an
app whose compose sets a real env var named after itself is untouched.
- Tools/hooks kept uppercase CFG_<TYPE>_ reads, so an instance
provisioned itself from the type's config and ignored its own values.
- Cloned hooks were never loaded: both loaders run at startup, before the
instance dir exists, so _appCallHook's `declare -F` found nothing and
every <slug>_install_* hook silently no-opped — for bookstack that is
the readiness probe and the admin bootstrap. Source the instance's own
scripts in-process, then regen arrays + manifest for later runs.
- bookstack's hook hardcoded the container name after `docker exec -e ...`
flags, where the rewriter can't see it, so an instance's admin bootstrap
ran against the BASE app's container — including a tinker DELETE of a
user. Target "$app_name" instead, and teach the rewriter the
container="<type>" assignment form used by auth adapters.
network_resources uniqueness:
UNIQUE(resource_type, resource_value) is right for 'ip' and 'port' but the
port-tag writer stores descriptive rows in the same table with INSERT OR
REPLACE, so every install DELETED the matching row from whichever app held
it. traefik_managed and url_accessible are booleans, so the whole table
could only ever hold one row of each. Observed live: installing a second
bookstack took all four traefik_managed/url_accessible rows from stoat and
bookstack, and removing that instance took the stolen rows with it.
Replace it with a partial unique index scoped to ip/port, and migrate
existing databases in place (SQLite can't drop a constraint, so the table
is rebuilt inside a transaction). The migration is invoked from
portUpdateComposeTags, not just databaseCreateTables — the latter only
runs from startPreInstall, which a working install never re-runs.
Verified: two bookstacks now hold port_tag_internal=80, traefik_managed
and url_accessible simultaneously; duplicate host ports and IPs are still
rejected; instance installs, serves HTTP 200, provisions its own admin in
its own database, and removes cleanly with no orphan rows.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Bookstack: create-admin fails on a reinstall because the account is
already provisioned. That took the generic-failure branch, which printed
the upstream defaults (admin@admin.com / password) as "the" login — those
credentials were replaced on the first install, so the one line a user
would act on was the wrong one. Detect "already exists" and say the
existing account was kept and its password not reset.
Firewall: uninstall deleted only resource_type='port', orphaning the
port_tag_*/traefik_managed/url_accessible rows the rebuild reads. Every
rebuild then walked ports for long-gone apps and printed "Skipped: <app>
(app not found)" per row. Widen the uninstall delete to all non-'ip' rows
(the source), and prune already-orphaned rows in the rebuild (the
self-heal). Pruning requires both no container dir AND status != 1 in the
apps table, so a mid-flight install can't prune itself.
Verified on a live bookstack reinstall: admin path reports correctly,
firewall pruned nextcloud + stalwart once, second run silent, 11 rules
added / 0 failed throughout.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Installing rocketchat failed with
invalid IPv4 address: ParseAddr("IP_DATA_2"): unable to parse IP
ipUpdateComposeTags allocates one IP per SERVICE_TAG_N annotation and fills
IP_TAG_i only where SERVICE_TAG_i exists. The four new apps tagged only their
primary service, so every sidecar — matrix's postgres, mattermost's postgres,
rocketchat's mongo, and fifteen of stoat's sixteen — kept a literal IP_DATA_n
in the deployed compose and docker refused to create the container.
Tag every service that carries an ipv4_address, index-aligned with its IP_TAG.
For stoat that also meant moving caddy from SERVICE_TAG_1 to _6 so the indices
line up with the IPs rather than the reading order.
mastodon had the same latent break (IP_TAG_2 and _3 untagged) and is fixed the
same way — it would have failed on first install for the same reason.
SERVICE_TAG carries the compose *key*, not container_name: 'libreportal app
restart <app> <service>' passes it to 'docker compose restart', which only
understands keys.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
CFG_NETWORK_MTU=1500 was baked blindly into the rootless daemon uplink
(DOCKERD_ROOTLESS_ROOTLESSKIT_MTU) and every app's container network. On links
whose real path MTU is below 1500 (Qubes/NAT/VPN with PMTU discovery blocked),
image manifests + tiny images pull fine but large image LAYERS stall and reset
mid-blob with "httpReadSeeker: ... EOF" — apps silently fail to install. Probed
here: path MTU ~1328, Docker at 1500 → black hole.
- New scripts/network/network_mtu.sh: networkDetectMtu (don't-fragment ICMP
ladder → largest standard MTU that gets through, 1500 when ICMP gives no
signal) + networkEffectiveMtu (CFG_NETWORK_MTU: a number is verbatim, "auto"
probes once and caches $docker_dir/.network_mtu) + networkRedetectMtu.
- CFG_NETWORK_MTU default 1500 -> auto. Rootless setup now writes the resolved
MTU into the override (and re-detects per install); the per-app NETWORK_MTU_TAG
uses the resolved value too. Explicit numbers still win.
Verified on this box: auto -> 1300; alpine (multi-MB) + the full ~100MB navidrome
image now pull to completion where they previously EOF'd. Applied live (override
1300, config=auto, cache=1300, rootless docker restarted).
Signed-off-by: librelad <librelad@digitalangels.vip>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Foundations for network-drift healing:
- ipInSubnet(ip, cidr): prefix-aware CIDR membership (pure bash), so
stored IPs can be checked against docker's real subnet. Honours the
actual prefix, so a healthy /16-subnet + /24-ip-range install is not
mistaken for drift.
- dockerInstallApp now accepts reset_network="ip": re-roll the static IP
from the current subnet but PRESERVE published host ports (clears only
IP rows; LIBREPORTAL_RESET_IP_ONLY keeps port_allocate reusing existing
ports). This is the heal path — a subnet move strands the IP, not the
port, so we don't churn bookmarks/forwards/proxy upstreams. reset="true"
still re-rolls both.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Two latent issues uncovered while designing network-drift detection:
- adoptDockerSubnet's comment claimed apps' IPs stay inside docker's
subnet after adoption. False: IPs are pinned to the old subnet's first
three octets, so adopting a different /24 base strands every app IP
out-of-subnet. Document the real behaviour + the heal paths.
- ipAllocation fell through from the existing-row branch to the
unconditional INSERT, which would violate UNIQUE(app,type,service).
Unreachable on today's reset path (rows are deleted first) but a hazard
for any direct caller; add an explicit return after reuse/reset.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
'local result=$(cmd)' resets $? to 0 (the local builtin's own exit), so the
following checkSuccess always saw success regardless of cmd's real exit — the
mechanism that masked the de-sudo write failures. Split declaration from
assignment ('local result; result=$(cmd)') across all 235 active-code sites
(84 files) so the command's exit reaches checkSuccess. No behaviour change
beyond $? now being accurate (no set -e in runtime code; multi-line
assignments transform safely).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Last app-specific bits out of central infra (from the per-app audit):
- traefik middleware: replace the hardcoded onlyoffice/owncloud exclude-list +
onlyoffice-headers special-case (in traefik_middlewares.sh AND
traefik_port_middlewares.sh) with two per-app hooks an app ships in
containers/<app>/scripts/<app>_traefik.sh:
appTraefikSkipsDefaultMiddleware_<app> (marker: opt out of default@file)
appTraefikExtraMiddlewares_<app> (echo extra middleware entries)
onlyoffice defines both; owncloud defines the skip marker. Two narrow hooks
(not one clever one) so behavior — incl. the different onlyoffice-headers
ordering between the two files — is preserved exactly. Verified with stubs:
identical middleware strings across normal/onlyoffice/owncloud × authelia/wl.
- moneyapp: add a placeholder icon (geometric banknote SVG, 512x512) so it no
longer falls back to default.svg in the WebUI.
Central traefik/compose code is now app-agnostic.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Two more runtime root file-primitive subsystems moved behind self-
validating root-owned helpers so the scoped sudoers needn't grant blanket
sudo sed/tee/cp on /etc (which is root-equivalent — sudo arg wildcards
match across '/', so even path-scoped entries are bypassable):
- scripts/system/libreportal-dns: {clear|add <ip>} — edits /etc/resolv.conf
only, validates the IP argument
- scripts/system/libreportal-ssh-access: authorized_keys + sshd
PasswordAuthentication management, with the lockout guards moved INTO the
helper (the trust boundary) so a compromised manager can't bypass them
- run_privileged: _runRootHelper dispatcher + runResolv / runSshAccess
(runOwnership now uses it too)
- init.sh: initRootHelpers installs all three helpers root:root 0755 with
the manager name baked in
- setup_dns -> runResolv (+ ping de-sudo'd, works unprivileged); host_access
+ webui_ssh_access -> runSshAccess
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Wireguard standalone touches /etc/wireguard + sysctl exclusively (genuine
root) -> runSystem for all its mkdir/chmod/sed/rm/grep/tee/qrencode. Traefik
dynamic configs live under containers/traefik (docker-install-owned) ->
runFileOp/runFileWrite (whitelist.yml, protectionauth.yml, the router-rewrite
awk|tee|mv in port_subdomains). sudo -u drops left.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
app_generate operates on the manager-owned install template -> runInstallOp
(cp/mv/sed); drop sudo on the interactive editor. localDnsApplyPihole edits
containers/pihole/.../custom.list (docker-install-owned) -> read via runFileOp,
build in a manager /tmp scratch, write back via runFileWrite.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Model A prototype (run start.sh AS the manager, escalate only via helpers):
- check_root.sh: accept the manager user, not root-only (init.sh keeps its own
install-time root check).
- init.sh: guard the top-level root-check + installer entrypoint with
BASH_SOURCE!=$0 so it runs ONLY when init.sh is executed directly; when
start.sh sources it as the manager the entrypoint (and its root check) no
longer fires.
Also: convert bare daemon-touching 'docker' calls (no helper -> hit the
nonexistent /var/run socket in rootless) to runFileOp docker across
app_status, app_health_*, network_prune, ip_is_available, check_docker_network,
backup_db (db dumps) and crontab_check_processor. cd&&compose rooted-branches
and 'docker compose --version' checks left as-is (rooted-only / no daemon).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
The apps SQLite DB ($docker_dir/$db_file) is owned by the manager user, so
read/write it AS the manager via runInstallOp instead of sudo (root). 48 call
sites across 28 scripts. In rooted this drops root->manager (correct owner);
in rootless it's the manager too (using runFileOp/dockerinstall here was the
'unable to open database' bug). The broken 'command -v sudo sqlite3' check
lines are left untouched (separate pre-existing issue).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
The firewall rebuild chose ufw-docker vs ufw from $EUID -eq 0 (am I root?)
rather than the docker mode. During a rootless install everything runs as
root, so it wrongly picked ufw-docker — which manages the rooted daemon's
DOCKER-USER chain that rootless never creates — and failed with 'Docker
instance libreportal doesn't exist'. (It was also inconsistent at runtime: the
non-root cron refresh always fell through to plain ufw.) Select by
CFG_DOCKER_INSTALL_TYPE so rootless always uses plain ufw (ports are published
on the host) and rooted always uses ufw-docker.
Also: ufw-docker needs the container name, not the app name — pass
service_name (e.g. libreportal-service) with an app_name fallback; route the
traefik-detect docker ps through runFileOp (was raw docker -> /var/run in
rootless); and move the ufw/ufw-docker sudo calls to runSystem.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Reinstall test on Debian 12 surfaced three rootless-only breakages (rooted
was byte-identical/fine):
1. pasta blocked by Debian's passt AppArmor profile (DENIED ptrace read ->
can't open container netns -> rootless dockerd never starts). Default
CFG_ROOTLESS_NET back to slirp4netns (reliable); pasta stays selectable
for hosts that relax the profile.
2. de-sudo mis-assigned helpers by owner. /docker management layer (apps DB
chowned to libreportal by install_sqlite, /docker/logs) is MANAGER-owned,
not dockerinstall. Add runInstallWrite; move apps-DB sqlite3 -> runInstallOp
and /docker/logs appends -> runInstallWrite. Revert ownership-SETUP scripts
(libreportal_folders, app_folder) to runSystem — they must run as root to
establish ownership during install. Container files (/docker/containers/<app>)
stay runFileOp.
3. kernel hardening sysctls written to /etc/sysctl/99-custom.conf, which
'sysctl --system' does not read -> never applied. Write them to
/etc/sysctl.d/99-libreportal-hardening.conf instead.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
FIX: dockerCommandRun rooted path is 'sudo $command' (unquoted word-split),
so 'docker ps --format "{{.Names}}"' was passing the format with LITERAL
quotes -> docker emitted '<name>' and the downstream grep never matched
(broken in rooted too). Switch all docker invocations to runFileOp, which
preserves args via "$@" in both modes (and runs as dockerinstall against
the rootless socket). Fixed monitoring.sh, dashy, tags_processor_network_mode.
Convert: jitsimeet (rm/wget/unzip/mv/sed/tee/gen-passwords on /docker ->
runFileOp/runFileWrite), authelia (config sed/mkdir/chmod/chown/secrets tee
-> runFileOp/runFileWrite; docker exec -> runFileOp docker, preserving
--password), reset_git (cp->/root runSystem, install-dir chown runInstallOp;
kept sudo -u manager). check_update/update_git_check need no change (all
sudo -u manager git, already least-privilege).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
firewall_initial_setup + firewall_clear_rules (ufw/ufw-docker),
host_access.sh (sshd/-T/-t, /etc/ssh, authorized_keys, systemctl reload),
set_socket_permissions (docker socket test/chmod), and webui_install_systemd
(systemd unit tee + systemctl) -> runSystem. These stay real-root in both
modes and define part of the eventual scoped allowlist. Left the
'sudo -u <manager> crontab' run-as-manager lines for a dedicated pass.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
All operate on /docker data-plane (DB at $docker_dir, compose files,
task dir /docker/.../frontend/data/tasks): sqlite3/find/sed/mkdir/chmod/
chown/mv/rm/mkfifo/truncate/install/tee -> runFileOp/runFileWrite. The
two systemctl enable/start calls in the check processor -> runSystem.
Dropped spurious sudo on text-only echo/grep/date in db_app_scan.
Byte-identical in rooted.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Give dockerCommandRunInstallUser an --argv mode that execs arguments
verbatim (sudo -u <user> env ... "$@") instead of bash -c "$*", and
point runFileOp at it. The old $*+bash -c re-parse silently mangled
backslashes/quotes in args — e.g. sed scripts (\1, \( become 1, ( ) and
the sqlite3 .backup arg — so rootless data-plane ops with regex were
broken. Verified: the WG_DEFAULT_DNS sed now applies correctly as the
install user. All existing runFileOp callers pass plain commands, so the
switch is safe (and fixes the latent sqlite3 case).
Convert scripts/network/dns/setup_dns.sh: /etc/resolv.conf edits and
ping -> runSystem; the WG_DEFAULT_DNS compose-file sed -> runFileOp.
Byte-identical in rooted; correct in rootless.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
The static per-app CFG_<APP>_HOST_NAME is gone. host_setup (the app's
canonical FQDN, feeding the legacy single DOMAINSUBNAME_DATA used by app env
vars, the app URL and trusted-domains) is now derived from the app's primary
Traefik port's subdomain: first recommended port, else first Traefik port;
@/root -> apex, set -> sub.domain, empty -> app-name. Removes HOST_NAME from
all app configs, the config-form field mapping (Hostname), the dead
headscale stub, and wireguard.sh (now uses host_setup). Completes the move to
dynamic per-port subdomain routing.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
setupLocalDnsRewrites points every configured domain at the server's LAN IP
inside the self-hosted resolver, so app subdomains resolve locally and hit
Traefik directly (valid certs, no router hairpin). AdGuard gets a wildcard
rewrite per domain via its REST API; Pi-hole gets per-host A records in the
supported, mounted custom.list (no wildcard support there). Safe by
construction: idempotent, guarded by installed-checks, cannot corrupt the
resolver. Hooked into the Apply-DNS actions and resolver install. Also drops
the dead HOST_NAME read from the setupDNSIP stub.
NOTE: needs a live smoke-test — the AdGuard API call and Pi-hole reload
can't be exercised without the running containers.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
An empty subdomain previously resolved to the domain apex, which would
collide on the root for any unconfigured Traefik port. Treat empty as the
app-name default (matching legacy behaviour); apex is reachable only via the
explicit @ / root sentinel.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Replace the static one-host-per-app model with per-port routers: each
Traefik-managed port carries a subdomain (12-col PORT format) and gets a
DOMAINSUBNAME_TAG_<n> host, so one container can serve unlimited hosts.
tagsProcessorPortSubdomains stamps per-port hosts (subdomain @/empty = apex,
multi-level allowed); tagsProcessorPortRouterBlocks comments out
# TRAEFIK_PORT_<n>_BEGIN/END blocks for non-Traefik ports so unfilled
placeholders never ship (mirrors GLUETUN_OFF). Convert all 27 router apps
(subdomains seeded from HOST_NAME; headscale admin. prefix -> subdomain).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
HOST_NAME was read but ignored — the FQDN was built from app_name, so 8
apps (vault, cloud, search, notes, social, meet, board, bookmark) routed at
the wrong host and Traefik disagreed with DNS. Build host_setup from
HOST_NAME (falling back to app_name); treat HOST_NAME="@"/"root" as the
domain apex (root-of-domain hosting, previously impossible). Document @ in
the Hostname field tooltip.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
A free, open, self-hosted app platform (GNU AGPLv3): one-click app deploys,
Traefik reverse proxy with automatic SSL, rootless Docker support, gluetun
VPN routing, and a web dashboard to manage it all.
Free & open forever to self-host; optional paid hosted services fund it.
See PROMISE.md.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
