"Cloud assets fail silently under backend load after publish"
▸ SYMPTOM
- Game fine in editor; after publish, players see floating ERROR meshes /
error.vmdl. - Prefab references appear rewritten to
error.vmdl. - No helpful retry: assets just missing under backend load.
▸ CAUSE
Cloud assets can fail silently under sbox.game backend load. There is no auto-retry. The engine may rewrite prefab refs to error.vmdl, baking the failure into what players load.
This is distinct from loose local resources not being listed (loose-resource-files-publish) and from whitelist API failures: same "worked in editor" shape, different fix.
▸ FIX
- Minimize cloud-asset dependence for launch-critical meshes/materials: prefer project-local packaged assets when you can.
- After every publish, verify each critical cloud asset loads in a player join (not only editor play).
- If placeholders appear: editor restart + republish.
- Heed the Publish Wizard's per-cloud-asset eligibility/license check (also relevant to play-fund-explained).
Minimize cloud-asset use, or pin and verify each one after publish: the documented reliable workaround remains restart + republish.
▸ WHY IT WORKS
Editor play often hits warm/local caches; published clients hit the cloud backend under shared load. Without retry, a single failed fetch becomes a permanent ERROR placeholder for that session, and prefab rewrites make the failure sticky until you republish clean references.
- Aligned pin-and-verify advice to source; removed needs-verification hedge.