Operation

This page sums up what an institution needs to know about running Moliri, before it starts and while it works with it.

In the browser

Moliri runs in the browser. Nothing is installed, neither on a workstation nor on a tablet.

Browser	Status
Edge, Chrome, Firefox, Safari	Supported
Desktop and iPad	The intended devices
Smartphone	Not intended

The interface is available in 7 languages: German, English, Spanish, French, Greek (Ελληνικά), Japanese and Ukrainian. Each person chooses their language on the right of the main menu.

Your own instance

Each institution gets an instance of its own with its own database. Data from different institutions never sits in the same database. Within the instance you separate areas through projects and permissions, see Permission model .

Moliri does not provide for several institutions or a collection hierarchy in one instance.

Moliri sets up the instance. After that, an administrator at your institution takes over its upkeep with the tools in this section.

Backup

What	How
Database	Hourly backup
Files (images, documents, 3D models)	Separately from the database, by their own procedure

Independently of this, every single change can be traced and undone in the history , including a whole import or a whole bulk edit. The history does not restore the state as of a particular date.

Media

Moliri only hands out originals and preview images after a permission check: to logged-in accounts with read permission on the file. Someone who only knows the address of an image does not get it. The exception is the images of publicly shared objects, see Shares .

Moliri prepares new images as an IIIF image service (Image API 3.0, level 0): one directory per image with tiles and sizes up to the size of the image itself. Images from before that still have the older derivatives. The image viewer continues to show them, but they are only available as IIIF after they have been prepared again. On an instance on your own premises, this happens in the instance’s container:

SH
php api/_func/iiif_backfill.php            # report: how many files have which derivatives
php api/_func/iiif_backfill.php --queue    # prepare every file without an IIIF directory again
php api/_func/ld_projection.php --full     # afterwards: rewrite manifests and public images

The preparation runs in the upload queue and removes a file’s old derivatives as soon as its IIIF directory is in place.

Linked Data and IIIF

Linked Data and IIIF are on as soon as the instance has a base URI . Two regular runs keep the public data up to date:

Run	When	What
Projection	Every minute	Rewrites the documents of publicly shared objects after changes, deletes what is no longer public and maintains the change feed
Full dump	Every night at 2:30	Writes `…/ld/dump` as JSON-LD and N-Triples

The instance sets up both runs when it starts, and only if a base URI is set. When the base URI is set for the first time, documents therefore only appear after the instance has been restarted. Without a base URI, the runs are dropped at the next start.

An instance on your own premises needs the forwarding rules for /id/, /ld/ and /iiif/ from the bundled nginx configuration, and a media service from version 1.33.0.

Monitoring

The instances are monitored. The monitoring reports faults to Moliri automatically.

Hosting

Location	Remark
Data centre in Germany	The usual case
Server on your own premises	Possible if your institution does not want to, or may not, let its data leave the building

The schema validation for LIDO exports runs in the browser and needs no internet access. Here too, the data does not leave the building.

Support

Questions, bug reports and requests go by email to [email protected] . There is also a chat in the application. It has to be switched on for each account individually (in the profile under Chat Support or in Permissions › Users) and is voluntary. The chat is subject to the terms of the provider Crisp, which the profile links to.