Disabling userland-proxy makes rootless dockerd require br_netfilter
(/proc/sys/net/bridge/bridge-nf-call-iptables), absent in the rootless
netns on Debian -> default bridge creation fails -> daemon won't start.
Drop the daemon.json userland-proxy=false write. Source-IP is preserved
at L7 by Traefik (X-Forwarded-For), so no real loss.
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>
- restic_install, crowdsec_update/verify_firewall/fix_priority: pure host
ops (apt/cscli/nft/systemctl, /etc/crowdsec) -> runSystem.
- kopia_backup/borg_restore: ignore-file/target tee+chown+mkdir -> runFileOp/
runFileWrite; kept the 'sudo -E -u dockerinstall' engine calls as-is —
those already run as the unprivileged backup user (least-privilege; the
scoped sudoers will permit (dockerinstall)).
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>
- New runInstallOp helper for manager install-dir/template ops (rooted:
sudo; rootless: run as the current manager user, which owns the tree).
- adguard.sh, traefik.sh: container-config sed -> runFileOp.
- crowdsec.sh: host crowdsec systemctl/apt-get -> runSystem.
- dashy_update_conf.sh: conf-file mkdir/chown/md5sum/tee -> runFileOp/
runFileWrite; docker ps/restart -> dockerCommandRun.
Deferred (cross-owner copy / temp-file across /tmp<->/docker, need rootless
env to bridge correctly): owncloud_setup_config.sh, adguard_auth.sh.
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>
Add -a/--append to runFileWrite so the pervasive /docker/logs log-append
idiom (`… | sudo tee -a $logs_dir/$docker_log_file`) routes through the
mode-aware helper instead of raw sudo.
Convert scripts/config/docker/docker_config_to_container.sh fully: all
ops target /docker app config + logs (data-plane), so md5sum/grep/chmod/
cmp/editor -> runFileOp and the log-appends -> runFileWrite -a.
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>
Enabling unprivileged user namespaces for rootless widens the kernel
attack surface reachable by unprivileged users (a known source of LPE
CVEs). Pair it with three distro-portable, low-impact sysctls that close
the surfaces those exploit chains rely on: kernel.kptr_restrict=2 (hide
kernel pointers), kernel.yama.ptrace_scope=1 (block cross-process
ptrace), net.core.bpf_jit_harden=2 (harden the JIT). Added as a separate
guarded LIBREPORTAL KERNEL HARDENING block so it's clearly deliberate and
independently idempotent.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Split the rootless network-driver knob out of network_docker into a
dedicated advanced subcategory (configs/network/network_rootless), marked
**ADVANCED** at both the file-header and field level so it's hidden behind
the advanced toggle in the WebUI. Registered in the network .category
order after network_docker. Pure config relocation — the install script
reads CFG_ROOTLESS_NET as a sourced var, so no script change needed.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Switch the rootless network stack from slirp4netns+builtin to pasta+
implicit (faster and propagates the real client source IP). The earlier
pasta+builtin attempt bricked the daemon because rootlesskit rejects
mismatched net/port-driver pairs; expose a single CFG_ROOTLESS_NET knob
(pasta default, slirp4netns fallback) and derive the matching port
driver in-script so an invalid combo can't be configured. Disable
userland-proxy in the rootless daemon.json (merged, not clobbered) so
containers see the real source IP. Both driver binaries are always
installed, so switching is a config flip + rootless re-setup.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
In rootless the container was set to user: <host-uid> (e.g. 1002:1002) with
group_add: <host-sub-gid> (e.g. 166528). Inside the daemon's user namespace
those are out of range — group_add made runc call setgroups() with an unmapped
GID ('setgroups: invalid argument'), so the WebUI container never started.
In rootless the container now runs as 0:0 (userns-root == the install user, which
owns the bind-mounts and the rootless socket) with socket gid 0. Rooted is
unchanged. Verified: libreportal-service comes up and talks to the rootless
socket.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
The rootless dockerd override forced NET=pasta + PORT_DRIVER=builtin, which
rootlesskit rejects ('pasta requires port driver none or implicit'), so the
daemon failed to start every time (the real cause behind 'rootless socket not
found'). Use slirp4netns + builtin (valid, still skips the userspace
port-handler). Verified: daemon now comes up, docker Server 29.5.2 responds.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
dockerCommandRunInstallUser ssh'd to <user>@localhost, but nothing set up an
SSH server/keys/authorized_keys, so every rootless setup command (daemon
install, systemctl --user) silently no-op'd. Replace with 'sudo -u <user> env
…' that sets XDG_RUNTIME_DIR / DBUS_SESSION_BUS_ADDRESS / DOCKER_HOST / PATH
explicitly; linger keeps the user systemd + /run/user/<uid> alive so
systemctl --user works. No SSH server, no keys, less attack surface, and
sudo -u to an unprivileged user is not a root escalation.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
useradd was missing its login-name argument (and -m), so it failed — silently,
because local result=$(...) swallowed the exit code and checkSuccess reported
success. The rootless install user was therefore never created, which cascaded
into 'invalid user dockerinstall' and a daemon that never came up. Pass the
username + -m (subordinate uid/gid ranges come from login.defs), unmasked.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
setupConfigsFromRepo / sync_configs_from_install used 'cp -a' of the template
over /docker/configs, so any fast/full deploy (which runs init.sh) silently
reset user config to template defaults — e.g. it flipped a live rooted box to
the new rootless template default and broke it. Use 'cp -an' (no-clobber):
fresh installs still get the full template, existing installs keep their values,
and new keys are still added by the add-only reconcile pass. This is also what
makes a rootless template default safe for existing rooted boxes.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Convert the backup/restore data-plane sudo calls (mkdir/chown/rm/sqlite3/tar/
gzip|tee) to runFileOp/runFileWrite. Rooted behaviour is identical (helper runs
sudo); rootless will run them as the unprivileged install user. Pilot subsystem
for the wider de-sudo. verify.sh's /tmp scratch ops left as-is (different
ownership domain, handled separately).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Single place that decides how a privileged op runs by Docker mode:
- runFileOp / runFileWrite: /docker data-plane ops — rooted uses sudo (identical
to today), rootless runs as the unprivileged install user (no root).
- runSystem: genuine system-admin ops, sudo in both modes, funnelled here so it
can later be confined to a scoped sudoers allowlist.
Call sites converted to these are byte-for-byte unchanged under rooted, so
existing/live boxes can't regress; rootless gets the de-privileged path.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
The SSH-access feature's files_ssh.sh array was never registered in
files_source.sh, leaving it unsourced and blocking the deploy auto-merge.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
init.sh appended 'libreportal ALL=(ALL) NOPASSWD: ALL' straight to /etc/sudoers
— a malformed line there locks out sudo entirely. Move it to a validated
/etc/sudoers.d/libreportal drop-in (visudo -cf before install, 0440 root:root).
The grant is still broad; this is the single managed file we tighten to a
scoped command allowlist once the runtime no longer needs broad root. Only runs
at install, so existing boxes are untouched.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Containers now run unprivileged by default — a container breakout maps to a
sub-UID, not host root. Rooted remains available as a legacy opt-in. Existing
installs keep their current mode (config reconciliation is add-only); fresh
installs get rootless. The rootless path already handles unprivileged ports
(ip_unprivileged_port_start=0) and userns.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
Convert the remaining sections off the legacy ?= query form to clean paths,
matching the Admin area:
/apps/<category> (was /apps?=<category>)
/app/<name>?tab=&task= (was /app?=<name>&tab=&task=)
/tasks/<category>?task= (was /tasks?=<category>&task=)
/backup/<tab> (was /backup?=<tab>)
Builders updated everywhere (sidebar, dashboard, notifications, tasks, apps,
app tabs, task-actions, setup watcher); parsers now read the resource from the
path with the legacy ?= kept as a fallback so old links/bookmarks still work
(server already serves index.html at any depth). Route table gains /apps* and
orders it before /app* (since '/apps' startsWith '/app'); active-nav and
config/apps data-loading recognise the new paths.
Tab/task remain ordinary query params (modifiers, not the primary resource).
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
