Webstar/LibrePortal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(install): bridge cross-owner folder copies + writable install log

Browse Source 
Two pre-existing bugs a genuinely-clean rootless install exposes:

copyFolder picked the copy user by destination only: a manager-owned
source (e.g. the install dir) copied into the dockerinstall-owned
containers/ ran the cp AS dockerinstall, which can't read the source ->
"cp: Permission denied". The `local result=$(...)` then masked the
failure (local returns 0) so checkSuccess printed success. This broke
installLibrePortalImageWebUI: the WebUI dir wasn't populated, so
initializeAppVariables couldn't read libreportal.config ("No app name
provided"), compose tags were never substituted, and the WebUI container
couldn't start (user: "USER_DATA"). Fix: when source and destination
owners differ (manager -> container), bridge with a tar pipe — the
manager reads, dockerinstall writes — with pipefail so a read-side
failure is no longer masked.

start.sh created the per-run install log with `sudo touch` (root:root
644) but tee's to it as the manager -> "tee: Permission denied" -> every
install-*.log was empty. Fix: chown the log to the user running the
install so the tee can append.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Signed-off-by: librelad <librelad@digitalangels.vip>
This commit is contained in:
librelad 2026-05-24 21:20:48 +01:00
parent fba5456218
commit a978749ee8
2 changed files with 23 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
scripts/function/folder/copy_folder.sh
View File

@ -8,9 +8,23 @@ copyFolder()
local user_name="$3" # advisory — the destination path determines the owner
# Write as the destination's owner — no root, no chown (see copyFile).
local op="runInstallOp"
[[ "$save_dir" == "$containers_dir"* || "$save_dir" == /docker/containers/* ]] && op="runFileOp"
local result=$($op cp -rf "$folder" "$save_dir")
if [[ "$save_dir" == "$containers_dir"* || "$save_dir" == /docker/containers/* ]]; then
if [[ "$folder" == "$containers_dir"* || "$folder" == /docker/containers/* ]]; then
# container -> container: same owner (dockerinstall), a plain cp works.
local result=$(runFileOp cp -rf "$folder" "$save_dir")
else
# Cross-owner: a manager-owned source (e.g. the install dir) into the
# container-owned destination. Under rootless a single cp can't read
# one end and write the other, so bridge with a tar pipe — the manager
# reads the source, dockerinstall writes the destination. No root, no
# chown; files land owned by the destination user. pipefail so a
# read-side failure isn't masked by a "successful" empty extract.
( set -o pipefail
runInstallOp tar -C "$(dirname "$folder")" -cf - "$folder_name" \
| runFileOp tar -C "$save_dir" -xf - )
fi
else
local result=$(runInstallOp cp -rf "$folder" "$save_dir")
fi
checkSuccess "Copying $folder_name to $save_dir"
}

5
start.sh
View File

@ -49,6 +49,11 @@ initLibrePortal()
install_log_path="/docker/logs/install-$(date +%Y%m%d-%H%M%S).log"
sudo mkdir -p /docker/logs 2>/dev/null
sudo touch "$install_log_path" 2>/dev/null
# Own it by whoever runs the install (the manager under Model A) so the
# tee below — which runs as that user, not root — can append. A
# sudo-touched file is root:root 644, leaving the manager unable to
# write it → the empty install-*.log files.
sudo chown "$(id -un):$(id -gn)" "$install_log_path" 2>/dev/null
sudo chmod 644 "$install_log_path" 2>/dev/null
export install_log_path
exec > >(tee -a "$install_log_path") 2>&1