duplicate-request-cache · shipping · also Duplicate reply cache, DRC

Duplicate request cache

A duplicate request cache retains request identities and completion information to recognize retries without repeating completed work.

Related Observatory article: Duplicate request cache on hesela.com

NFSv3 servers can return a remembered completion status for a recognized duplicate. Cache loss or entry reuse can remove that protection, including after reconnecting over a reliable transport. [1]

NFSv4.1 uses session slots and sequence IDs. Omitting a full reply does not remove execution tracking: a retry can return NFS4ERR_RETRY_UNCACHED_REP. Persistent sessions have additional restart obligations. [2]

This is request history, not a file-data cache or a content-deduplication index. Duplicate detection alone does not establish that the file's bytes are durably stored. [1]

Relations

related

Retry deduplication, Idempotency, NFS, NAS, NFS write verifier

Citations

1. RFC 1813: NFSv3 duplicate request cache, section 4.5. (accessed 2026-10-19)
2. RFC 8881: NFSv4.1 session replay, optional reply caching and persistence, section 2.10.6. (accessed 2026-10-19)

Further reading

3. Storage timeouts: an expired timer is not a cancelled write (accessed 2026-10-19)

JSON: /data/terms/duplicate-request-cache.json