Refactors many install scripts to use the shared `nginx_enable_site` helper instead of manual symlink/default-site/reload sequences, making webserver setup more consistent and less error-prone. It also replaces hardcoded PHP-FPM socket paths with `get_php_fpm_socket` (including templated substitutions where needed) across nginx and caddy configs to improve compatibility with varying PHP versions and socket locations.
* UmlautAdaptarr: use the Debian 13 Microsoft repo with its 2025 signing key
The container is Debian 13 (var_version 13) but the script configures the
Microsoft debian/12 prod repo with suite bookworm.
The move to debian/13 was reverted in #8392 because apt reported "OpenPGP
signature verification failed" (#8385), diagnosed at the time as Microsoft
not having populated the trixie repo. The actual cause was the signing key:
debian/13/prod is signed by EE4D7792F748182B (microsoft-2025.asc), while the
script kept microsoft.asc (EB3E94ADBE1229CF). The suite and URL were changed
but the key was not.
The trixie repo carries the required packages at current patch level
(dotnet-sdk-8.0 8.0.423-1, aspnetcore-runtime-8.0 8.0.29-1).
This makes the call identical to the six other scripts already on the
Debian 13 Microsoft repo (technitiumdns, igotify, mail-archiver, rdtclient,
fileflows, immichframe).
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* UmlautAdaptarr: migrate existing containers to the Debian 13 repo on update
Per review feedback: the install-side fix only helps new containers.
Existing ones keep the Debian 12 repo indefinitely, because update_script
only fetches a GitHub release and never touches apt.
Repoints them when the stale Debian 12 repo is detected, mirroring the
conditional setup_deb822_repo calls in ct/technitiumdns.sh and
ct/fileflows.sh. The guard makes this a no-op once migrated.
setup_deb822_repo's microsoft* globs also clear the pre-#9839
microsoft-prod.sources layout; its key lived in /usr/share/keyrings,
which cleanup_old_repo_files does not cover, so it is removed explicitly
as ct/technitiumdns.sh does.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
---------
Co-authored-by: Tim Moore <tim.moore@ingenuity.com.au>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
* fix(bazarr): store data in /var/lib/bazarr instead of inside /opt/bazarr
The installer creates /var/lib/bazarr and update_script() uses it as the
"is Bazarr installed" guard, but nothing pointed Bazarr at it. Bazarr
defaults its data directory to <install dir>/data, resolved from its own
source file rather than WorkingDirectory, so a stock container kept
config/ db/ backup/ cache/ log/ restore/ in /opt/bazarr/data - the
directory fetch_and_deploy_gh_release redeploys over - while
/var/lib/bazarr stayed empty.
Pass -c /var/lib/bazarr in the unit, matching the -data= flag the other
*arr scripts use. bazarr.py re-execs its child with sys.argv[1:], so the
flag survives the restart Bazarr performs while loading its config.
update_script() gains a one-time migration for existing installs. It runs
outside the release check so containers already on the latest version are
migrated too, is skipped when the unit already names a data directory,
refuses (before stopping the service) when both locations hold data, and
copies before removing so a failed migration leaves the original database
in place.
Fixes#16097
* fix(bazarr): drop explanatory comments per review
Both scripts provision Debian 13 containers (var_version 13) but add a
Debian 12 (bookworm) non-free apt source to obtain unrar.
Because the base image only enables the `main` component, trixie's
non-free is never enabled, so bookworm is the only source offering
unrar and the containers deterministically install the Debian 12 build
(1:6.2.6-1+deb12u1) instead of trixie's 1:7.1.8-1.
Mylar3 additionally never removes the source, leaving bookworm
non-free and non-free-firmware permanently enabled at priority 500 on
a Debian 13 system.
Aligns both with install/sabnzbd-install.sh, which already sets up the
same unrar dependency correctly against trixie.
Co-authored-by: Tim Moore <tim.moore@ingenuity.com.au>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
* Pin Vikunja install/update to v2.3.0
Set both install and update scripts to deploy Vikunja v2.3.0 instead of latest. The update flow now checks against this pinned release and includes context that v2.4.0 is temporarily avoided due to an upstream startup failure caused by the packaged systemd SystemCallFilter.
* minor fix
The container was switched from bookworm to trixie in #16036, but
software-properties-common does not exist in Debian 13. apt aborts the
whole dependency line on the single missing package (exit code 100), so
a fresh install never gets past "Installing Dependencies".
The package only provides add-apt-repository, which this script never
calls — no PPAs or custom repos are configured anywhere. It is a dead
dependency, so dropping it needs no replacement; every other package on
that line exists in trixie.
Fixes#16053
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Install Java via setup_java instead of distro openjdk
Apache Tika's Java runtime was installed by calling apt directly for
openjdk-17-jre-headless. This tied the JDK to whatever Debian ships, and
that coupling has now broken: openjdk-17 is no longer available on Debian
13 (Trixie), where openjdk-21 is current. It also made this the only
script in install/ that installs Java by hand, while every other Java
application in the repository uses the shared setup_java helper.
Switch to setup_java, which installs Eclipse Temurin from the Adoptium
repository. Temurin is versioned independently of the Debian release, so
a future distribution upgrade cannot remove the JDK out from under the
application the way this one did.
Java 21 satisfies both current and upcoming Tika: 3.x requires Java 11 or
newer, and 4.x raises the floor to Java 17.
The surrounding msg_info/msg_ok pair is dropped because setup_java emits
its own progress messages; keeping them would nest the output.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Bump Apache-Tika container to Debian 13
The container declared Debian 12 (Bookworm) as its base, which had become
an outlier: 475 of the repository's ct scripts target Debian 13 and only
22 still target 12. build.func also sets trixie as the default ARM64
template codename, so this script's declared base had drifted away from
the repository default.
The declared version is not cosmetic. check_container_os_guard() reads
var_os/var_version as the script's recommended OS and compares them
against a container's /etc/os-release, prompting on interactive runs and
aborting headless ones when they disagree. Debian 12 was also the reason
the previous openjdk-17 dependency looked valid; on Debian 13 that
package no longer exists, which is addressed in the preceding commit.
Existing Debian 12 containers keep working, because update_script only
replaces the Tika jar and never touches the operating system or Java.
They will, however, now hit the OS mismatch guard on update and be asked
to move to Debian 13 first. That is intended rather than a regression,
but it is a visible change for existing users.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Migrate Termix database to the new db/data layout introduced in 2.5.1, fail fast on backend build errors, and precreate the db/data directory on fresh installs.
Co-authored-by: P3RF3CTION <13474113+P3RF3CTION@users.noreply.github.com>
Overwriting /etc/mongod.conf with only the replication block wiped
out storage.dbPath from the mongodb-org package default, so mongod
fell back to its compiled-in /data/db path, which was never created.
Append the replication section instead of replacing the file.
* Fix Nametag update failing with missing tailwindcss module
npm ci during update ran with devDependencies omitted, likely because
NODE_ENV=production was already present in the environment by then
(npm's documented default: omit=dev when NODE_ENV=production). Since
@tailwindcss/postcss is a devDependency, the subsequent `npm run build`
failed with "Cannot find module '@tailwindcss/postcss'". Force devDeps
to be installed regardless of NODE_ENV via --include=dev.
* Apply --include=dev to fresh installs too, for parity with update fix
Couldn't confirm within this repo why devDependencies would only be
omitted on update and not on a fresh install - create_backup/
restore_backup are plain file copies and nothing in tools.func sets
NODE_ENV or writes an .npmrc. Since npm ci runs before .env is sourced
in both scripts, apply the same explicit --include=dev to the installer
as defense-in-depth against the same class of failure.
* RomM: use backup helpers in update / clear folder
Replace manual .env backup/restore with create_backup and restore_backup helpers, and add CLEAN_INSTALL=1 flag to fetch_and_deploy_gh_release.
* add screenscraper env
* add rom_patcher
* add ruffle and emulatorJS
Define `path.data` and `path.logs` in the Zammad install script's Elasticsearch config so the service uses the expected data and log directories when running in the single-node local setup.
Update both `ct/omada.sh` and `install/omada-install.sh` to parse the Omada version directly from the `_v..._linux` segment of the Debian filename. This removes the old dependency on a timestamp suffix format and ensures `.omada` gets a valid version when TP-Link package naming varies.
The /data/plugins symlink fails in default unprivileged containers with Operation not permitted. Link plugins via /opt/wanderer/source/db/data/plugins instead so PluginDir() can discover them without root-level symlinks.
Fixes#15799
* tools.func: replace rm -rf with find for safer directory cleanup
* tools.func: replace rm -rf with find for safer directory cleanup
* refactor(tools.func): centralize deploy tail + trap-based tmpdir cleanup
Extract the duplicated post-download logic of the fetch_and_deploy_*
helpers into two shared functions and switch per-branch cleanup to a
single RETURN trap.
- _deploy_source_tarball: shared source-tarball tail (6 call sites)
- _deploy_unpacked_archive: shared prebuild-archive tail (3 call sites)
- RETURN trap per function guarantees tmpdir/unpack_tmp cleanup on every
return path, replacing ~40 manual `rm -rf "$tmpdir"` lines
- CLEAN_INSTALL now lives in the helpers instead of 12 copies
Behavior-preserving except: codeberg prebuild gains .txz support and
uses helper return codes; from_url resets shopt on error paths.
* Add _download_source_tarball with retry/validation
Introduce `_download_source_tarball` helper that validates gzip integrity after download and retries up to 3 times. This guards against truncated-but-valid-HTTP responses from GitHub/GitLab/Codeberg on-the-fly archive generation. Replace ad-hoc curl/curl_download calls in fetch_and_deploy_* functions with the new helper. Also remove redundant tmpdir cleanup before early returns (tmpdir is cleaned up at function exit).
* Fix Manyfold Ruby setup home initialization
Create the `manyfold` service user before calling `setup_ruby` so `/home/manyfold` exists when rbenv profile snippets are written. Also harden `setup_ruby` by creating `$HOME` if missing, preventing profile-write failures for installers that pass a home directory before creating the user.
* Update install/manyfold-install.sh
---------
Co-authored-by: Sam Heinz <sam@samheinz.com>
PR #14818 replaced the setup wizard curl with CLI commands but omitted
the schema migration step, causing fresh installs to fail when creating
the admin user. Trigger schema init via HTTP after config-ini.
Fixes#15828
MickLesk