1개 문서가 "dig-store-cache" 태그에 분류되었습니다

The normative content-replication flywheel: how a DIG Node turns every read into a new holder — connect over mTLS, discover holders through the Kademlia provider DHT (find_providers) and the real-time signed holdings-announce (gossip opcode 222), fetch and merkle-verify the bytes against the chain-anchored root, admit the capsule into the bounded on-disk LRU cache (1 GiB default, pins exempt), then announce the gain so the reader becomes a discoverable holder — plus the eviction retract that keeps the holder map honest, the first-party landing gate that stops a stranger deciding what this node stores, and the untrusted-peer property that makes the loop safe: content is accepted because it verifies, never because of who served it.