Enterprise companion to HypAware: receives logs forwarded from HypAware instances across an org, spools them per gateway with fsync-before-ack durability, moves them into the kernel's Iceberg cache, archives closed days to Iceberg on S3 or local disk, and serves admin SQL spanning cache + archive.
Design rationale lives in the numbered LLP documents under
llp/; source files carry
@ref LLP NNNN#anchor annotations back into them.
npm install # needs ../hypaware checked out (file: dependency)
HYPSERVER_IDENTITY_SECRET=... HYPSERVER_ADMIN_TOKEN=... npm startConfiguration is entirely env-driven — see src/config.js for every
knob (port, data dir, S3 archive, retention, backpressure, caps).
Secrets only ever arrive via environment.
npm run smokeBoots the real daemon against a temp dir and walks every plane: enrollment policy tokens, ingest durability + idempotency, dataset registration + schema skew, backpressure, revocation, capped admin SQL, archive export, ack-coupled eviction, the cache+archive query union, and restart durability.
hypaware-server-admin wraps every admin endpoint for docker-exec
workflows (mint-token, list-gateways, revoke-gateway, query,
run-archive, …). Auth via HYPSERVER_ADMIN_TOKEN.
See the header of Dockerfile — the build context must
contain both hypaware/ and hypaware-server/ side by side.