root 8b92518d21 G1P: vectord persistence to storaged + scrum (3 fixes incl. 3-way convergent)
Adds optional persistence to vectord (G1's HNSW vector search). Single-
file framed format per index — eliminates the torn-write class that
the 3-way convergent scrum finding identified:

  _vectors/<name>.lhv1  — single binary blob:
      [4 bytes magic "LHV1"]
      [4 bytes envelope_len uint32 BE]
      [envelope bytes — JSON params + metadata + version]
      [graph bytes — raw hnsw.Graph.Export]

Pre-extraction: internal/catalogd/store_client.go → internal/storeclient/
shared package, since both catalogd and vectord need it. Same pattern as
the pre-D5 catalogclient extraction.

Optional via [vectord].storaged_url config (empty = ephemeral mode).
On startup: List + Load each persisted index. After Create / batch Add /
DELETE: Save (or Delete from storaged). Save failures are logged-not-
fatal — in-memory state is the source of truth in flight.

Acceptance smoke G1P 8/8 PASS — kill+restart preserves state, post-
restart search returns dist=0 (graph round-trips exactly), DELETE
removes the file, post-delete restart shows count=0.

All 8 smokes (D1-D6 + G1 + G1P) PASS deterministically. The g1_smoke
gained scripts/g1_smoke.toml that disables persistence so the
in-memory API test stays decoupled from any rehydrate-from-storaged
state contamination.

Cross-lineage scrum on shipped code:
  - Opus 4.7 (opencode):                     1 BLOCK + 5 WARN + 3 INFO
  - Kimi K2-0905 (openrouter):               1 BLOCK + 2 WARN
  - Qwen3-coder (openrouter):                2 BLOCK + 2 WARN + 1 INFO

Fixed (3 — 1 convergent + 2 single-reviewer):
  C1 (Opus + Kimi + Qwen 3-WAY CONVERGENT WARN): Save was non-atomic
    across two PUTs — envelope-succeeds + graph-fails left a half-
    saved index that passed the "both present" List filter and
    silently mismatched metadata against vectors on Load.
    Fix: collapse to single framed file (no torn-write window
    possible).
  O-B1 (Opus BLOCK): isNotFound substring-matched "key not found"
    against the wrapped error message — brittle, any 5xx body
    containing that text would silently misclassify as missing.
    Fix: errors.Is(err, storeclient.ErrKeyNotFound).
  O-I3 (Opus INFO): handleAdd pre-validation only covered id+dim;
    NaN/Inf/zero-norm could still fail mid-batch leaving partial
    commits. Fix: extend pre-validation to call ValidateVector
    (newly exported) per item before any commit.

Dismissed (3 false positives):
  K-B1 + Q-B1 ("safeKey double-escapes %2F segments") — false
    convergent. Wire-protocol escape is decoded by storaged's chi
    router on the way in; on-disk key is the original literal.
    %2F round-trips correctly through PathEscape → URL → chi decode
    → S3 key.
  Q-B2 ("List vulnerable to race conditions") — vectord is single-
    process; no concurrent Save against List in the same vectord.

Deferred (3): rehydrate per-index timeout (G2+ multi-index scale),
saveAfter request ctx (matches G0 timeout deferral), Encode RLock
during slow writer (documented as buffer-only API).

The C1 finding is the strongest signal of the cross-lineage filter:
three independent reviewers all flagged the same torn-write hazard.
Single-file framing eliminates the class — there's now no Persistor
state where envelope and graph can disagree.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-29 01:33:23 -05:00

168 lines
5.9 KiB
Bash
Executable File

#!/usr/bin/env bash
# G1P smoke — vectord persistence across restart.
#
# Validates:
# - Create index + add 50 vectors → Save fires (storaged shows
# _vectors/persist_demo.json AND _vectors/persist_demo.hnsw)
# - Search w-001 → recall=1 with distance ≈ 0
# - Kill vectord → restart → initial Refresh enumerates persisted
# indexes, Loads each back into the registry
# - Same search post-restart → still recall=1 (state survived)
# - DELETE → both files removed from storaged → restart → index gone
#
# Requires storaged + vectord both up.
#
# Usage: ./scripts/g1p_smoke.sh
set -euo pipefail
cd "$(dirname "$0")/.."
export PATH="$PATH:/usr/local/go/bin"
echo "[g1p-smoke] building storaged + vectord + gateway..."
go build -o bin/ ./cmd/storaged ./cmd/vectord ./cmd/gateway
pkill -f "bin/(storaged|vectord|gateway)" 2>/dev/null || true
sleep 0.3
PIDS=()
TMP="$(mktemp -d)"
cleanup() {
echo "[g1p-smoke] cleanup"
for p in "${PIDS[@]}"; do [ -n "$p" ] && kill "$p" 2>/dev/null || true; done
rm -rf "$TMP"
}
trap cleanup EXIT INT TERM
poll_health() {
local port="$1" deadline=$(($(date +%s) + 5))
while [ "$(date +%s)" -lt "$deadline" ]; do
if curl -sS --max-time 1 "http://127.0.0.1:$port/health" >/dev/null 2>&1; then return 0; fi
sleep 0.05
done
return 1
}
NAME="persist_demo"
DIM=8
echo "[g1p-smoke] launching storaged..."
./bin/storaged > /tmp/storaged.log 2>&1 &
PIDS+=($!)
poll_health 3211 || { echo "storaged failed"; tail /tmp/storaged.log; exit 1; }
# Clean any prior persisted state for this index name.
curl -sS -o /dev/null -X DELETE "http://127.0.0.1:3211/storage/delete/_vectors/$NAME.lhv1" || true
launch_vectord() {
./bin/vectord > /tmp/vectord.log 2>&1 &
VECTORD_PID=$!
PIDS+=($VECTORD_PID)
poll_health 3215 || { echo "vectord failed"; tail /tmp/vectord.log; return 1; }
}
launch_gateway() {
./bin/gateway > /tmp/gateway.log 2>&1 &
PIDS+=($!)
poll_health 3110 || { echo "gateway failed"; tail /tmp/gateway.log; return 1; }
}
echo "[g1p-smoke] launching vectord (round 1) → gateway..."
launch_vectord
launch_gateway
FAILED=0
echo "[g1p-smoke] create index + add 50 vectors:"
HTTP="$(curl -sS -o /dev/null -w '%{http_code}' -X POST http://127.0.0.1:3110/v1/vectors/index \
-H 'Content-Type: application/json' \
-d "{\"name\":\"$NAME\",\"dimension\":$DIM,\"distance\":\"cosine\"}")"
if [ "$HTTP" != "201" ]; then echo " ✗ create → $HTTP"; FAILED=1; fi
# Build a 50-vector batch.
python3 - <<EOF > "$TMP/batch.json"
import json
items = []
for i in range(50):
v = [0.0] * $DIM
v[i % $DIM] = 1.0
v[(i+1) % $DIM] = (i % 13) * 0.01
items.append({"id": f"w-{i:03d}", "vector": v, "metadata": {"row": i}})
print(json.dumps({"items": items}))
EOF
LEN="$(curl -sS -X POST "http://127.0.0.1:3110/v1/vectors/index/$NAME/add" \
-H 'Content-Type: application/json' \
-d @"$TMP/batch.json" | jq -r '.length')"
if [ "$LEN" = "50" ]; then echo " ✓ added 50 → length=50"; else echo " ✗ add → length=$LEN"; FAILED=1; fi
echo "[g1p-smoke] verify storaged has the persistence file:"
FILE_KEY="_vectors/$NAME.lhv1"
FILE_HTTP="$(curl -sS -o /dev/null -w '%{http_code}' "http://127.0.0.1:3211/storage/get/$FILE_KEY")"
if [ "$FILE_HTTP" = "200" ]; then
echo "$FILE_KEY present in storaged"
else
echo " ✗ file=$FILE_HTTP"; FAILED=1
fi
echo "[g1p-smoke] search pre-restart:"
python3 - <<EOF > "$TMP/q.json"
import json
i = 1
v = [0.0] * $DIM
v[i % $DIM] = 1.0
v[(i+1) % $DIM] = (i % 13) * 0.01
print(json.dumps({"vector": v, "k": 1}))
EOF
TOP_PRE="$(curl -sS -X POST "http://127.0.0.1:3110/v1/vectors/index/$NAME/search" \
-H 'Content-Type: application/json' -d @"$TMP/q.json" | jq -r '.results[0].id')"
if [ "$TOP_PRE" = "w-001" ]; then echo " ✓ pre-restart top hit = w-001"; else echo " ✗ top=$TOP_PRE"; FAILED=1; fi
echo "[g1p-smoke] kill + restart vectord (rehydrate path):"
kill $VECTORD_PID 2>/dev/null || true
wait $VECTORD_PID 2>/dev/null || true
sleep 0.3
launch_vectord
sleep 0.2 # let initial Refresh complete (no API to query rehydration done)
echo "[g1p-smoke] vectord rehydrated index list shows $NAME:"
NAMES_HTTP="$(curl -sS http://127.0.0.1:3110/v1/vectors/index | jq -r '.names | length')"
if [ "$NAMES_HTTP" = "1" ]; then echo " ✓ list count=1 after restart"; else echo " ✗ count=$NAMES_HTTP"; FAILED=1; fi
INFO_LEN="$(curl -sS "http://127.0.0.1:3110/v1/vectors/index/$NAME" | jq -r '.length')"
if [ "$INFO_LEN" = "50" ]; then echo " ✓ length=50 after restart (state survived)"; else echo " ✗ length=$INFO_LEN"; FAILED=1; fi
echo "[g1p-smoke] search post-restart:"
TOP_POST="$(curl -sS -X POST "http://127.0.0.1:3110/v1/vectors/index/$NAME/search" \
-H 'Content-Type: application/json' -d @"$TMP/q.json" | jq -r '.results[0].id')"
DIST_POST="$(curl -sS -X POST "http://127.0.0.1:3110/v1/vectors/index/$NAME/search" \
-H 'Content-Type: application/json' -d @"$TMP/q.json" | jq -r '.results[0].distance')"
if [ "$TOP_POST" = "w-001" ]; then
echo " ✓ post-restart top hit = w-001 (dist=$DIST_POST)"
else
echo " ✗ post-restart top=$TOP_POST"; FAILED=1
fi
echo "[g1p-smoke] DELETE then restart → index gone:"
curl -sS -o /dev/null -X DELETE "http://127.0.0.1:3110/v1/vectors/index/$NAME"
FILE_HTTP="$(curl -sS -o /dev/null -w '%{http_code}' "http://127.0.0.1:3211/storage/get/$FILE_KEY")"
if [ "$FILE_HTTP" = "404" ]; then
echo " ✓ persistence file removed from storaged"
else
echo " ✗ file=$FILE_HTTP after delete"; FAILED=1
fi
kill $VECTORD_PID 2>/dev/null || true
wait $VECTORD_PID 2>/dev/null || true
sleep 0.3
launch_vectord
sleep 0.2
NAMES_HTTP="$(curl -sS http://127.0.0.1:3110/v1/vectors/index | jq -r '.names | length')"
if [ "$NAMES_HTTP" = "0" ]; then echo " ✓ post-delete restart list count=0"; else echo " ✗ count=$NAMES_HTTP"; FAILED=1; fi
if [ "$FAILED" -eq 0 ]; then
echo "[g1p-smoke] G1P acceptance gate: PASSED"
exit 0
else
echo "[g1p-smoke] G1P acceptance gate: FAILED"
exit 1
fi