baraba/Baradb
1
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(persist): unnamed FTS/HNSW indexes survive restart (nameless replay DDL)

Browse Source
This commit is contained in:
dimgigov
2026-07-30 16:27:36 +03:00
parent 214b9cf346
commit 821b668d87
2 changed files with 79 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/barabadb/query/executor.nim
+12 -2
View File
@@ -1321,7 +1321,13 @@ proc executeQueryImpl(ctx: ExecutionContext, astNode: Node, params: seq[WireValu
ctx.ftsIndexes[colKey] = ftsIdx
# Persist reconstructed DDL so restoreEngines can rebuild the index
# from table data after a restart (replay re-writes the same key).
let ftsDdl = "CREATE INDEX " & idxName & " ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING FTS"
# Unnamed indexes: the colKey fallback name is dotted ("docs.content")
# and unparseable, so persist the nameless form — replay regenerates
# the same colKey default.
let ftsDdl = if stmt.ciName.len > 0:
"CREATE INDEX " & idxName & " ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING FTS"
else:
"CREATE INDEX ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING FTS"
ctx.db.put(SchemaFtsIndexPrefix & colKey, cast[seq[byte]](ftsDdl))
return okResult(msg="CREATE INDEX " & idxName & " on " & stmt.ciTarget & " USING FTS")
@@ -1359,7 +1365,11 @@ proc executeQueryImpl(ctx: ExecutionContext, astNode: Node, params: seq[WireValu
ctx.vectorIndexes[colKey] = hnswIdx
# Persist reconstructed DDL so restoreEngines can rebuild the index
# from table data after a restart (replay re-writes the same key).
let vecDdl = "CREATE INDEX " & idxName & " ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING HNSW"
# Unnamed indexes: persist the nameless form (see FTS branch above).
let vecDdl = if stmt.ciName.len > 0:
"CREATE INDEX " & idxName & " ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING HNSW"
else:
"CREATE INDEX ON " & stmt.ciTarget & " (" & stmt.ciColumns.join(", ") & ") USING HNSW"
ctx.db.put(SchemaVecIndexPrefix & colKey, cast[seq[byte]](vecDdl))
return okResult(msg="CREATE INDEX " & idxName & " on " & stmt.ciTarget & " USING HNSW")
tests/test_schema_persist.nim
+67
View File
@@ -168,6 +168,73 @@ suite "Schema persistence":
db2.close()
removeDir(dir)
test "Unnamed FTS index survives reopen":
let dir = "/tmp/baradb_schema_persist_fts_noname"
removeDir(dir)
block:
var db = newLSMTree(dir)
var ctx = newExecutionContext(db)
check execSql(ctx, "CREATE TABLE docs (id INTEGER PRIMARY KEY, content TEXT)").success
check execSql(ctx, "INSERT INTO docs (id, content) VALUES (1, 'quick brown fox')").success
# No index name — idxName defaults to colKey (docs.content) at execution
check execSql(ctx, "CREATE INDEX ON docs (content) USING FTS").success
check ctx.ftsIndexes.hasKey("docs.content")
db.close()
# Reopen fresh context (simulates process restart)
block:
var db2 = newLSMTree(dir)
var ctx2 = newExecutionContext(db2)
# Persisted DDL must be replayable — the dotted fallback name is not.
check ctx2.ftsIndexes.hasKey("docs.content")
if ctx2.ftsIndexes.hasKey("docs.content"):
check ctx2.ftsIndexes["docs.content"].search("quick", limit = 10).len >= 1
# Nameless persisted DDL must not confuse DROP INDEX name matching
let d = execSql(ctx2, "DROP INDEX content")
check d.success
check "docs.content" notin ctx2.ftsIndexes
let (found, _) = db2.get(SchemaFtsIndexPrefix & "docs.content")
check not found
db2.close()
block:
var db3 = newLSMTree(dir)
var ctx3 = newExecutionContext(db3)
check "docs.content" notin ctx3.ftsIndexes # no ghost rebuild
db3.close()
removeDir(dir)
test "Unnamed HNSW index survives reopen":
let dir = "/tmp/baradb_schema_persist_vec_noname"
removeDir(dir)
block:
var db = newLSMTree(dir)
var ctx = newExecutionContext(db)
check execSql(ctx, "CREATE TABLE vecs (id INTEGER PRIMARY KEY, embedding TEXT)").success
check execSql(ctx, "INSERT INTO vecs (id, embedding) VALUES (1, '[1.0, 0.0, 0.0]')").success
# No index name — idxName defaults to colKey (vecs.embedding)
check execSql(ctx, "CREATE INDEX ON vecs (embedding) USING HNSW").success
check ctx.vectorIndexes.hasKey("vecs.embedding")
db.close()
# Reopen fresh context (simulates process restart)
block:
var db2 = newLSMTree(dir)
var ctx2 = newExecutionContext(db2)
check ctx2.vectorIndexes.hasKey("vecs.embedding")
if ctx2.vectorIndexes.hasKey("vecs.embedding"):
check vengine.search(ctx2.vectorIndexes["vecs.embedding"],
@[1.0'f32, 0.0'f32, 0.0'f32], k = 5).len >= 1
let d = execSql(ctx2, "DROP INDEX embedding")
check d.success
check "vecs.embedding" notin ctx2.vectorIndexes
let (found, _) = db2.get(SchemaVecIndexPrefix & "vecs.embedding")
check not found
db2.close()
block:
var db3 = newLSMTree(dir)
var ctx3 = newExecutionContext(db3)
check "vecs.embedding" notin ctx3.vectorIndexes # no ghost rebuild
db3.close()
removeDir(dir)
test "Graph survives reopen":
let dir = "/tmp/baradb_schema_persist_graph"
removeDir(dir)