LibrePhotos

The current update_scan_counter sets job.failed = True after a single per-photo error and the flag is sticky for the rest of the run: job.failed = failed or job.failed So a scan that processes 152520 photos and errors on 4 of them — a real production case visible in LongRunningJob row 58b891a5-5776-4a27-9ae3-c5610a80ee27 from a user's instance, with progress 152520/152520 and result.status='partial_failure' — shows up as the red "Failed" banner in the admin UI even though 99.997% of work succeeded. result already correctly carries the partial_failure status and the per-photo error list; only the sticky boolean is wrong. This change replaces both occurrences of the sticky assign

Every backend → sidecar HTTP client (face_recognition, image_similarity, clip_embeddings, exif/metadata reader, image_captioning, llm, thumbnail, tags, plus the periodic health checker) calls requests.post/get/delete with no timeout. requests defaults to no timeout, which means a stalled sidecar — face_recognition wedged on a pathological image, exiftool hung mid-process, image_similarity in a tight loop — parks the caller on an infinite socket read. The caller is most often a long-running scan job running inside a single django-q2 task; the task ends up "running" forever in the admin UI with no progress, and the global Q_CLUSTER timeout is set to 10000000 (115 days) so the supervisor never

When I restart the backend container on my library (~400k photos, on the stable image), postgres sits at ~85–95% CPU for around 20 minutes while the backend itself is mostly idle. Watching the logs and `pg_stat_activity`, the load is coming from a sequence of paginated SELECTs over `Photo.clip_embeddings`, each feeding a `POST /build/` call into the image_similarity Flask service. Tracing it back: the backend entrypoint runs `python manage.py build_similarity_index`, which enqueues `build_image_similarity_index(user)` for every user. That function (in `api/image_similarity.py`) DELETEs the in-memory FAISS index in the Flask service and then re-streams every photo's `clip_embeddings` to it i

> ℹ️ **Note** > > This PR body was truncated due to platform limits. This PR contains the following updates: | Package | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | Type | Update | |---|---|---|---|---|---| | [@badeball/cypress-cucumber-preprocessor](https://redirect.github.com/badeball/cypress-cucumber-preprocessor) | [`^20.1.2` → `^24.0.0`](https://renovatebot.com/diffs/npm/@badeball%2fcypress-cucumber-preprocessor/20.1.2/24.0.1) |  | ![confidence](https://developer.mend.io/api/mc/ba

# 🛑 Before you create an issue make sure that: - Your issue is **strictly related to LibrePhotos** itself. Questions about setting up a reverse proxy belong in what ever reverse proxy you are using. - You have read the [documentation](https://docs.librephotos.com) thoroughly. - You have searched for a similar issue among all the former issues (even closed ones). - You have tried to replicate the issue with a clean install of the project. - You have asked for help on our Discord server [LibrePhotos](https://discord.gg/xwRvtSDGWb) if your issue involves general "how to" questions **When Submitting please remove every thing above this line** # 🐛 Bug Report * [ ] 📁 I've Included a ZIP file

> ℹ️ **Note** > > This PR body was truncated due to platform limits. This PR contains the following updates: | Package | Change | [Age](https://docs.renovatebot.com/merge-confidence/) | [Confidence](https://docs.renovatebot.com/merge-confidence/) | Type | Update | |---|---|---|---|---|---| | [@babel/core](https://babel.dev/docs/en/next/babel-core) ([source](https://redirect.github.com/babel/babel/tree/HEAD/packages/babel-core)) | [`7.24.9` → `7.29.0`](https://renovatebot.com/diffs/npm/@babel%2fcore/7.24.9/7.29.0) |  | ![confidence](https://developer.mend.io/api/mc/badges/confidence/npm/@babel%2f

**Describe the enhancement you'd like** Try catching known failures and giving users useful feedback from crashes with actionable items, either only in the logs, or if possible in the web interface as well. **Describe why this will benefit the LibrePhotos** This would help new users solve common issues when they happen, and would cut down the amount of support needed in the discord. **Additional context** Add any other context or screenshots about the enhancement request here.

We currently do not provide any feedback on what went wrong if a user has not set a directory. The program should check if the path is available. If not and the user is an admin, a modal should pop up, where you can set the path. If the path is not set, but it is a regular user, an error should pop, saying "There is currently no path set for this user. Please contact your admin to set up a path."

Currently, a couple of operations are not yet explained like: "full rescan", "train faces" and "faces rescan". We should add them to the frontend and make sure that it does not get too long or that we could hide it behind a question mark.

I've read some of the posts and reviews from a year ago on Google Photos replacements but I'm still confused about which self-hosted galleries meet the following criteria: - Points to photo folder and leaves photos in tact for easy backup and data portability. Albums are reflected in folder structure so if I switch galleries in the future, they will be retained. - Good experience on Android, securely available outside local network, but doesn't have to be an app - Support for video is a must - AI features not necessary - Ideal if it's available through Yunohost - Must load large directories with large numbers of photos and videos quickly - l would prefer to use a separate solution fo