Module pool 

HOP data pool: in-memory index backed by sharded on-disk storage.

On-disk layout

The pool root contains two subdirectories, blobs/ and meta/, each sharded into 256 subdirectories named 00–ff after the first byte of the content hash. An entry with hash H is stored as:

• blobs/<H[0:2]>/<H>.blob — raw payload bytes
• meta/<H[0:2]>/<H>.meta — SCALE-encoded HopEntryMeta

Recovery

On startup the pool scans every meta/ shard, decodes each .meta file, and rebuilds the in-memory index. .meta files that are corrupt, have an unexpected version, or lack a sibling .blob are deleted. Then the corresponding blobs/ shard is scanned and any .blob without an entry in the freshly-built index (orphan) is also deleted. Stale .tmp.* files left by a previous crash are removed during both scans.

Structs

HopDataPool

HOP data pool with disk-backed blob storage and in-memory metadata index.