Currently on the metadata indexer was implemented as one but it was dedicated to create one-shot tasks with an indirection on the scheduler.
This would:
- simplify the stack by removing moving parts (scheduler, storage access db for content indexer...).
- allow better monitoring (as we already have a grafana dashboard for journal clients)
- allow indexation to be retried [1] on error
- stop one index computation failure to fail the full batch indexation
Indexers: