docs: Update supported file types, provider management, connector management

[aimurphy]

• Update supported file types
• Add cloud connector sync
• Update cloud connector management
• Inspect failed tasks
• Expand model provider management and add more warnings about changing and removing models/providers
• Deleted the openrag-documentation.pdf since we now ingest from docs.openr.ag

Closes #1128

[aimurphy]

I don't think this is true. Please correct me if otherwise.

[Wallgau]

I think that statement is true.
src/services/search_service.py detects all indexed embedding_model values, generates a query embedding for each, then builds multiple KNN clauses (one per model field) and runs them together via dis_max. So with more embedding models, retrieval work increases (more embedding calls + larger query), which can make Chat retrieval slower, depending on provider latency and OpenSearch load.

[aimurphy]

Ok thanks! I will add it back.

Caveat: If you remove a provider, then the Chat cannot generate an embedding with the appropriate embedding model. What happens in that case? Will the Chat ignore documents embedded by the missing providing? Or does it search with a "fallback" embedding model (potentially incorrect/missing results because the embedding dimensions/structure don't match)?

[Wallgau]

from my understanding but always good to double check with @edwinjosechittilappilly.
If a provider is removed, OpenRAG still discovers embedding models from indexed documents and attempts to generate a query embedding for each, so if one of those models is no longer available, retrieval may fail and Chat may return an error or no useful answer; there is no safe cross-model fallback for those documents because embedding spaces and dimensions differ, and the system does not consistently fallback to searching only documents tied to still-available models.

[aimurphy]

Is it ok to delete this?

[aimurphy]

@Wallgau

[github-actions]

Build successful! ✅
Deploying docs draft.
Deploy successful! View draft

[mendonk]

One typo, approved

[mendonk]

Suggested change

	To update docs ingested from cloud storage connectors], see [Configure connectors](/knowledge-connectors).
	To update docs ingested from cloud storage connectors, see [Configure connectors](/knowledge-connectors).

[Wallgau]

Hey @aimurphy! Just synced with the team, @edwinjosechittilappilly is running tests to check files type. So let's hold on this PR until we can give you a solid list.
cc @edwinjosechittilappilly @prasanthcaibmcom