iemejia commented on code in PR #12400:
URL: https://github.com/apache/gluten/pull/12400#discussion_r3631331310
##########
backends-velox/src/main/scala/org/apache/gluten/config/VeloxConfig.scala:
##########
@@ -534,10 +535,38 @@ object VeloxConfig extends ConfigRegistry {
val COLUMNAR_VELOX_FILE_HANDLE_CACHE_ENABLED =
buildStaticConf("spark.gluten.sql.columnar.backend.velox.fileHandleCacheEnabled")
.doc(
- "Disables caching if false. File handle cache should be disabled " +
- "if files are mutable, i.e. file content may change while file path
stays the same.")
+ "Enables caching of open file handles to avoid repeated open/close
overhead. " +
+ "Benefits both local filesystems (fewer open/close syscalls and file
descriptor " +
+ "churn) and remote filesystems/object stores (reused connection
state). Should be " +
+ "disabled if files are mutable, i.e. file content may change while
the file path " +
+ "stays the same.")
.booleanConf
- .createWithDefault(false)
+ .createWithDefault(true)
+
+ val COLUMNAR_VELOX_NUM_CACHE_FILE_HANDLES =
+
buildStaticConf("spark.gluten.sql.columnar.backend.velox.numCacheFileHandles")
+ .doc(
+ "Maximum number of entries in the file handle cache. Each entry holds
an open " +
+ "file descriptor (local FS) or connection state (remote FS). Note
that on " +
+ "local filesystems, high values may approach the OS file descriptor
limit " +
+ "(ulimit -n). On remote object stores (S3, ABFS, GCS) entries
represent " +
+ "network connections/sockets rather than per-file OS file
descriptors, but " +
+ "they can still count toward OS resource limits (ulimit -n).")
+ .intConf
+ .checkValue(_ > 0, "must be a positive number")
+ .createWithDefault(10000)
Review Comment:
The file handle cache is a bounded LRU (`SimpleLRUCache`), so
`numCacheFileHandles` is an upper bound on retained handles, not a
preallocation — descriptors are only held for files actually touched and are
reclaimed by both LRU eviction and the TTL. The default of 10000 is
deliberately half of Velox's own default (20000, `HiveConfig.cpp`), so it is
already more conservative than upstream, and the config's docstring explicitly
documents the `ulimit -n` interaction. Environments with a low FD limit can
lower it via `spark.gluten.sql.columnar.backend.velox.numCacheFileHandles`.
##########
backends-velox/src/test/scala/org/apache/spark/sql/execution/VeloxFileHandleCacheSuite.scala:
##########
@@ -0,0 +1,366 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.spark.sql.execution
+
+import org.apache.gluten.config.VeloxConfig
+import org.apache.gluten.execution.{BasicScanExecTransformer,
VeloxWholeStageTransformerSuite}
+
+import org.apache.spark.SparkConf
+
+import java.io.FileNotFoundException
+import java.nio.file.NoSuchFileException
+
+/**
+ * Test suite for Velox file handle cache behavior.
+ *
+ * Tests correctness, config propagation, and edge cases for the file handle
cache which caches open
+ * file handles (descriptors) to avoid repeated open/close overhead.
+ */
+class VeloxFileHandleCacheSuite extends VeloxWholeStageTransformerSuite {
+ override protected val resourcePath: String = "/parquet-for-read"
+ override protected val fileFormat: String = "parquet"
+
+ // TTL for file handle cache eviction (used in sparkConf and sleep
calculations).
+ // Kept small to minimize CI time; the TTL test only asserts scan
correctness after
+ // the window elapses (it passes whether or not eviction has occurred), so a
short
+ // wait is sufficient and does not introduce flakiness.
+ private val ttlMs = 500
+ private val ttlWaitMs = ttlMs + 500 // TTL + buffer for lazy eviction on
next access
+
+ /** Walks the exception cause chain looking for an instance of the given
type. */
+ private def hasCauseOfType(e: Throwable, cls: Class[_ <: Throwable]):
Boolean = {
+ var cause = e.getCause
+ while (cause != null) {
+ if (cls.isInstance(cause)) return true
+ cause = cause.getCause
+ }
+ false
+ }
+
+ override protected def sparkConf: SparkConf = {
+ super.sparkConf
+ .set(VeloxConfig.COLUMNAR_VELOX_FILE_HANDLE_CACHE_ENABLED.key, "true")
+ .set(VeloxConfig.COLUMNAR_VELOX_FILE_HANDLE_EXPIRATION_DURATION_MS.key,
ttlMs.toString)
+ .set(VeloxConfig.COLUMNAR_VELOX_NUM_CACHE_FILE_HANDLES.key, "10000")
+ }
+
+ test("basic scan correctness with file handle cache enabled") {
+ // Verify that enabling file handle cache produces correct scan results
+ withTempPath {
+ dir =>
+ spark
+ .range(10000)
+ .selectExpr("id", "cast(id % 7 as int) as category", "id * 1.5 as
value")
+ .repartition(10)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val df = spark.read.parquet(dir.getCanonicalPath)
+ df.createOrReplaceTempView("t")
+
+ runQueryAndCompare("SELECT count(*) FROM t") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ runQueryAndCompare("SELECT sum(value) FROM t WHERE category = 3") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ runQueryAndCompare("SELECT category, count(*) FROM t GROUP BY
category") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ }
+ }
+
+ test("repeated scans produce consistent results") {
+ // Repeated scans of the same files must produce identical results
regardless
+ // of whether handles are served from cache or re-opened after TTL
eviction.
+ withTempPath {
+ dir =>
+ spark
+ .range(5000)
+ .selectExpr("id", "cast(id as string) as name")
+ .repartition(50) // 50 files to exercise many cache entries
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+ val expected = spark.read.parquet(path).count()
+ assert(expected == 5000)
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(path))
+
+ // Scan the same files multiple times - results must be consistent
+ for (i <- 1 to 5) {
+ val count = spark.read.parquet(path).count()
+ assert(
+ count == expected,
+ s"Iteration $i: expected $expected rows but got $count")
+ }
+
+ // Verify aggregation consistency across repeated scans
+ val firstSum =
spark.read.parquet(path).selectExpr("sum(id)").collect()(0).getLong(0)
+ for (i <- 1 to 3) {
+ val sum =
spark.read.parquet(path).selectExpr("sum(id)").collect()(0).getLong(0)
+ assert(
+ sum == firstSum,
+ s"Iteration $i: sum mismatch, expected $firstSum but got $sum")
+ }
+ }
+ }
+
+ test("many small files do not cause errors with file handle cache") {
+ // Verify that scanning many small files with caching enabled does not
cause
+ // file descriptor exhaustion or other resource-related errors.
+ withTempPath {
+ dir =>
+ // Create 200 small parquet files
+ spark
+ .range(20000)
+ .selectExpr("id", "uuid() as payload")
+ .repartition(200)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val fileCount = dir.listFiles().count(_.getName.endsWith(".parquet"))
+ assert(fileCount >= 200, s"Expected at least 200 files, got
$fileCount")
+
+ // Verify scans go through Gluten/Velox
+
checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(dir.getCanonicalPath))
+
+ // Scan all files - should work without resource errors
+ val count = spark.read.parquet(dir.getCanonicalPath).count()
+ assert(count == 20000)
+
+ // Scan again - results must remain consistent
+ val count2 = spark.read.parquet(dir.getCanonicalPath).count()
+ assert(count2 == 20000)
+ }
+ }
+
+ test("filtered scan correctness with file handle cache") {
+ // Verify that predicate pushdown works correctly with cached file handles.
+ // This exercises the row group skipping path through cached handles.
+ withTempPath {
+ dir =>
+ spark
+ .range(100000)
+ .selectExpr(
+ "id",
+ "cast(id % 10 as int) as partition_key",
+ "cast(id * 0.01 as double) as metric")
+ .repartition(20)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](
+ spark.read.parquet(path).where("partition_key = 5"))
+
+ // Filter that matches ~10% of rows
+ val filtered = spark.read.parquet(path).where("partition_key =
5").count()
+ assert(filtered == 10000, s"Expected 10000 filtered rows, got
$filtered")
+
+ // Range filter
+ val rangeFiltered = spark.read.parquet(path).where("id >=
50000").count()
+ assert(rangeFiltered == 50000, s"Expected 50000 range-filtered rows,
got $rangeFiltered")
+
+ // Re-run same filters - results must remain consistent
+ val filtered2 = spark.read.parquet(path).where("partition_key =
5").count()
+ assert(filtered2 == filtered, "Filtered count mismatch on repeated
scan")
+ }
+ }
+
+ test("scan after file deletion does not silently return wrong data") {
+ // If a file is deleted between scans, the next scan should either:
+ // - Succeed with the original count (cached FD keeps inode alive on Linux)
+ // - Succeed with a reduced count (deleted file not accessible)
+ // - Throw a file-not-found error
+ // The key invariant: it must NOT silently return incorrect data.
+ withTempPath {
+ dir =>
+ spark
+ .range(1000)
+ .selectExpr("id")
+ .repartition(5)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+ // First scan populates the cache
+ val count1 = spark.read.parquet(path).count()
+ assert(count1 == 1000)
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(path))
+
+ // Delete one parquet file
+ val parquetFiles =
dir.listFiles().filter(_.getName.endsWith(".parquet"))
+ assert(parquetFiles.nonEmpty)
+ val deletedFile = parquetFiles.head
+ val deletedRows =
spark.read.parquet(deletedFile.getCanonicalPath).count()
+ assert(deletedFile.delete(), s"Failed to delete
${deletedFile.getCanonicalPath}")
+
+ // On Linux, the cached FD to the deleted file may still work
(unlinked inode).
+ // Either way, the remaining files should be readable.
+ // The scan may also throw if the FS detects the missing file.
+ try {
+ val count2 = spark.read.parquet(path).count()
+ // The count should be either (count1 - deletedRows) or count1
+ // depending on whether the OS kept the inode accessible
+ assert(
+ count2 == count1 || count2 == count1 - deletedRows,
+ s"Unexpected count after deletion: $count2 (original: $count1,
deleted: $deletedRows)")
+ } catch {
+ case e: FileNotFoundException =>
+ // Direct file-not-found exception.
+ case e: NoSuchFileException =>
+ // NIO equivalent of FileNotFoundException.
+ case e: Exception
+ if hasCauseOfType(e, classOf[FileNotFoundException]) ||
+ hasCauseOfType(e, classOf[NoSuchFileException]) =>
+ // Wrapped file-not-found in the cause chain (e.g., SparkException
wrapping).
+ case e: Exception
+ if e.getMessage != null &&
+ (e.getMessage.contains("FileNotFoundException") ||
+ e.getMessage.contains("No such file") ||
+ e.getMessage.contains("Path does not exist") ||
+ e.getMessage.contains("does not exist")) =>
+ // Fallback: message-based matching for FS implementations that use
+ // custom exception types (e.g., Hadoop, Velox native errors).
+ }
+ }
+ }
+
+ test("scans remain correct after TTL expiration window") {
+ // Correctness guard: verify that scans produce correct results after the
+ // configured TTL (set in sparkConf) has elapsed and cached handles may
+ // have been evicted. This does NOT directly assert that eviction occurred
+ // (Velox exposes no JVM-visible eviction counter), but it exercises the
+ // re-open path: if a handle was evicted, the scan must transparently
+ // re-open the file and return the same data. Combined with the "scan after
+ // file deletion" test -- which proves cached handles keep the inode alive
--
+ // this gives reasonable coverage that the TTL wiring works end-to-end.
Review Comment:
The overwrite-and-observe approach isn't deterministic in this integration
test: Spark writes Parquet with unique `part-<uuid>.parquet` names and re-lists
the directory on every `spark.read.parquet`, so a rewrite produces new paths
the cache has never seen — the stale handle is simply never queried again
rather than serving old data. Combined with Linux inode semantics (an in-place
rewrite reuses the inode and is visible through the cached FD; an atomic rename
leaves the old FD on an unlinked inode) and the absence of any JVM-visible
eviction counter in Velox, there is no reliable way to force observably-stale
content. We instead added a best-effort eviction probe (delete a cached file,
wait past the TTL, re-scan) that drives the expire→reopen path and asserts the
invariant that must always hold: the scan never silently returns corrupted data.
##########
backends-velox/src/test/scala/org/apache/spark/sql/execution/VeloxFileHandleCacheSuite.scala:
##########
@@ -0,0 +1,366 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+package org.apache.spark.sql.execution
+
+import org.apache.gluten.config.VeloxConfig
+import org.apache.gluten.execution.{BasicScanExecTransformer,
VeloxWholeStageTransformerSuite}
+
+import org.apache.spark.SparkConf
+
+import java.io.FileNotFoundException
+import java.nio.file.NoSuchFileException
+
+/**
+ * Test suite for Velox file handle cache behavior.
+ *
+ * Tests correctness, config propagation, and edge cases for the file handle
cache which caches open
+ * file handles (descriptors) to avoid repeated open/close overhead.
+ */
+class VeloxFileHandleCacheSuite extends VeloxWholeStageTransformerSuite {
+ override protected val resourcePath: String = "/parquet-for-read"
+ override protected val fileFormat: String = "parquet"
+
+ // TTL for file handle cache eviction (used in sparkConf and sleep
calculations).
+ // Kept small to minimize CI time; the TTL test only asserts scan
correctness after
+ // the window elapses (it passes whether or not eviction has occurred), so a
short
+ // wait is sufficient and does not introduce flakiness.
+ private val ttlMs = 500
+ private val ttlWaitMs = ttlMs + 500 // TTL + buffer for lazy eviction on
next access
+
+ /** Walks the exception cause chain looking for an instance of the given
type. */
+ private def hasCauseOfType(e: Throwable, cls: Class[_ <: Throwable]):
Boolean = {
+ var cause = e.getCause
+ while (cause != null) {
+ if (cls.isInstance(cause)) return true
+ cause = cause.getCause
+ }
+ false
+ }
+
+ override protected def sparkConf: SparkConf = {
+ super.sparkConf
+ .set(VeloxConfig.COLUMNAR_VELOX_FILE_HANDLE_CACHE_ENABLED.key, "true")
+ .set(VeloxConfig.COLUMNAR_VELOX_FILE_HANDLE_EXPIRATION_DURATION_MS.key,
ttlMs.toString)
+ .set(VeloxConfig.COLUMNAR_VELOX_NUM_CACHE_FILE_HANDLES.key, "10000")
+ }
+
+ test("basic scan correctness with file handle cache enabled") {
+ // Verify that enabling file handle cache produces correct scan results
+ withTempPath {
+ dir =>
+ spark
+ .range(10000)
+ .selectExpr("id", "cast(id % 7 as int) as category", "id * 1.5 as
value")
+ .repartition(10)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val df = spark.read.parquet(dir.getCanonicalPath)
+ df.createOrReplaceTempView("t")
+
+ runQueryAndCompare("SELECT count(*) FROM t") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ runQueryAndCompare("SELECT sum(value) FROM t WHERE category = 3") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ runQueryAndCompare("SELECT category, count(*) FROM t GROUP BY
category") {
+ checkGlutenPlan[BasicScanExecTransformer]
+ }
+ }
+ }
+
+ test("repeated scans produce consistent results") {
+ // Repeated scans of the same files must produce identical results
regardless
+ // of whether handles are served from cache or re-opened after TTL
eviction.
+ withTempPath {
+ dir =>
+ spark
+ .range(5000)
+ .selectExpr("id", "cast(id as string) as name")
+ .repartition(50) // 50 files to exercise many cache entries
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+ val expected = spark.read.parquet(path).count()
+ assert(expected == 5000)
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(path))
+
+ // Scan the same files multiple times - results must be consistent
+ for (i <- 1 to 5) {
+ val count = spark.read.parquet(path).count()
+ assert(
+ count == expected,
+ s"Iteration $i: expected $expected rows but got $count")
+ }
+
+ // Verify aggregation consistency across repeated scans
+ val firstSum =
spark.read.parquet(path).selectExpr("sum(id)").collect()(0).getLong(0)
+ for (i <- 1 to 3) {
+ val sum =
spark.read.parquet(path).selectExpr("sum(id)").collect()(0).getLong(0)
+ assert(
+ sum == firstSum,
+ s"Iteration $i: sum mismatch, expected $firstSum but got $sum")
+ }
+ }
+ }
+
+ test("many small files do not cause errors with file handle cache") {
+ // Verify that scanning many small files with caching enabled does not
cause
+ // file descriptor exhaustion or other resource-related errors.
+ withTempPath {
+ dir =>
+ // Create 200 small parquet files
+ spark
+ .range(20000)
+ .selectExpr("id", "uuid() as payload")
+ .repartition(200)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val fileCount = dir.listFiles().count(_.getName.endsWith(".parquet"))
+ assert(fileCount >= 200, s"Expected at least 200 files, got
$fileCount")
+
+ // Verify scans go through Gluten/Velox
+
checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(dir.getCanonicalPath))
+
+ // Scan all files - should work without resource errors
+ val count = spark.read.parquet(dir.getCanonicalPath).count()
+ assert(count == 20000)
+
+ // Scan again - results must remain consistent
+ val count2 = spark.read.parquet(dir.getCanonicalPath).count()
+ assert(count2 == 20000)
+ }
+ }
+
+ test("filtered scan correctness with file handle cache") {
+ // Verify that predicate pushdown works correctly with cached file handles.
+ // This exercises the row group skipping path through cached handles.
+ withTempPath {
+ dir =>
+ spark
+ .range(100000)
+ .selectExpr(
+ "id",
+ "cast(id % 10 as int) as partition_key",
+ "cast(id * 0.01 as double) as metric")
+ .repartition(20)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](
+ spark.read.parquet(path).where("partition_key = 5"))
+
+ // Filter that matches ~10% of rows
+ val filtered = spark.read.parquet(path).where("partition_key =
5").count()
+ assert(filtered == 10000, s"Expected 10000 filtered rows, got
$filtered")
+
+ // Range filter
+ val rangeFiltered = spark.read.parquet(path).where("id >=
50000").count()
+ assert(rangeFiltered == 50000, s"Expected 50000 range-filtered rows,
got $rangeFiltered")
+
+ // Re-run same filters - results must remain consistent
+ val filtered2 = spark.read.parquet(path).where("partition_key =
5").count()
+ assert(filtered2 == filtered, "Filtered count mismatch on repeated
scan")
+ }
+ }
+
+ test("scan after file deletion does not silently return wrong data") {
+ // If a file is deleted between scans, the next scan should either:
+ // - Succeed with the original count (cached FD keeps inode alive on Linux)
+ // - Succeed with a reduced count (deleted file not accessible)
+ // - Throw a file-not-found error
+ // The key invariant: it must NOT silently return incorrect data.
+ withTempPath {
+ dir =>
+ spark
+ .range(1000)
+ .selectExpr("id")
+ .repartition(5)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+ // First scan populates the cache
+ val count1 = spark.read.parquet(path).count()
+ assert(count1 == 1000)
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(path))
+
+ // Delete one parquet file
+ val parquetFiles =
dir.listFiles().filter(_.getName.endsWith(".parquet"))
+ assert(parquetFiles.nonEmpty)
+ val deletedFile = parquetFiles.head
+ val deletedRows =
spark.read.parquet(deletedFile.getCanonicalPath).count()
+ assert(deletedFile.delete(), s"Failed to delete
${deletedFile.getCanonicalPath}")
+
+ // On Linux, the cached FD to the deleted file may still work
(unlinked inode).
+ // Either way, the remaining files should be readable.
+ // The scan may also throw if the FS detects the missing file.
+ try {
+ val count2 = spark.read.parquet(path).count()
+ // The count should be either (count1 - deletedRows) or count1
+ // depending on whether the OS kept the inode accessible
+ assert(
+ count2 == count1 || count2 == count1 - deletedRows,
+ s"Unexpected count after deletion: $count2 (original: $count1,
deleted: $deletedRows)")
+ } catch {
+ case e: FileNotFoundException =>
+ // Direct file-not-found exception.
+ case e: NoSuchFileException =>
+ // NIO equivalent of FileNotFoundException.
+ case e: Exception
+ if hasCauseOfType(e, classOf[FileNotFoundException]) ||
+ hasCauseOfType(e, classOf[NoSuchFileException]) =>
+ // Wrapped file-not-found in the cause chain (e.g., SparkException
wrapping).
+ case e: Exception
+ if e.getMessage != null &&
+ (e.getMessage.contains("FileNotFoundException") ||
+ e.getMessage.contains("No such file") ||
+ e.getMessage.contains("Path does not exist") ||
+ e.getMessage.contains("does not exist")) =>
+ // Fallback: message-based matching for FS implementations that use
+ // custom exception types (e.g., Hadoop, Velox native errors).
+ }
+ }
+ }
+
+ test("scans remain correct after TTL expiration window") {
+ // Correctness guard: verify that scans produce correct results after the
+ // configured TTL (set in sparkConf) has elapsed and cached handles may
+ // have been evicted. This does NOT directly assert that eviction occurred
+ // (Velox exposes no JVM-visible eviction counter), but it exercises the
+ // re-open path: if a handle was evicted, the scan must transparently
+ // re-open the file and return the same data. Combined with the "scan after
+ // file deletion" test -- which proves cached handles keep the inode alive
--
+ // this gives reasonable coverage that the TTL wiring works end-to-end.
+ withTempPath {
+ dir =>
+ spark
+ .range(5000)
+ .selectExpr("id", "id * 2 as doubled")
+ .repartition(20)
+ .write
+ .parquet(dir.getCanonicalPath)
+
+ val path = dir.getCanonicalPath
+
+ // First scan populates the cache
+ val count1 = spark.read.parquet(path).count()
+ assert(count1 == 5000)
+
+ // Verify scans go through Gluten/Velox
+ checkGlutenPlan[BasicScanExecTransformer](spark.read.parquet(path))
+
+ val sum1 =
spark.read.parquet(path).selectExpr("sum(id)").collect()(0).getLong(0)
+
+ // Wait for TTL to expire
+ Thread.sleep(ttlWaitMs)
Review Comment:
The sleep here is not a correctness gate. The assertions tolerate both
outcomes — handle still cached (full count) or evicted (reduced count /
file-not-found) — so clock jitter or late eviction cannot make the test flaky;
it only ever asserts the no-silent-corruption invariant. Bounded polling would
require an observable eviction signal to await, but Velox exposes none to the
JVM, so there is no condition to poll on. The wait is kept short (TTL 500 ms +
buffer) to minimize CI impact.
##########
docs/velox-configuration.md:
##########
@@ -30,7 +30,8 @@ nav_order: 16
| spark.gluten.sql.columnar.backend.velox.directorySizeGuess
| ⚓ Static | 32KB | Deprecated, rename to
spark.gluten.sql.columnar.backend.velox.footerEstimatedSize
|
| spark.gluten.sql.columnar.backend.velox.driverSideBroadcastHashTableBuild
| 🔄 Dynamic | false | Enable driver-side broadcast hash
table build. When enabled, the hash table is built and serialized on the
driver, then broadcast to executors. When disabled, each executor builds its
own hash table from the broadcast data.
|
| spark.gluten.sql.columnar.backend.velox.enableTimestampNtzValidation
| 🔄 Dynamic | false | Enable validation fallback for
TimestampNTZ type. When true, any plan containing TimestampNTZ will fall back
to Spark execution. When false, allows native execution for TimestampNTZ scan.
|
-| spark.gluten.sql.columnar.backend.velox.fileHandleCacheEnabled
| ⚓ Static | false | Disables caching if false. File
handle cache should be disabled if files are mutable, i.e. file content may
change while file path stays the same.
|
+| spark.gluten.sql.columnar.backend.velox.fileHandleCacheEnabled
| ⚓ Static | true | Enables caching of open file handles
to avoid repeated open/close overhead. Benefits both local filesystems (fewer
open/close syscalls and file descriptor churn) and remote filesystems/object
stores (reused connection state). Should be disabled if files are mutable, i.e.
file content may change while the file path stays the same.
|
+| spark.gluten.sql.columnar.backend.velox.fileHandleExpirationDurationMs
| ⚓ Static | 10m | Expiration time for cached file
handles. Handles not accessed within this duration are evicted from the cache.
This prevents stale handles from accumulating (e.g., expired HDFS leases,
closed remote connections). Accepts a Spark duration string (e.g., "10m",
"600s") or a plain number interpreted as milliseconds. A value of 0 disables
TTL-based eviction.
|
Review Comment:
The only user-facing default is this docs table, which now reads `10m`
(following the earlier review suggestion to prefer a Spark-style literal over
`600000ms`). The `600000` is the native C++ fallback constant
(`kVeloxFileHandleExpirationDurationMsDefault` in `VeloxConfig.h`), which must
be a numeric `int64` and is only used if the value isn't supplied by the JVM.
In practice `GlutenConfigUtil.parseConfig` always normalizes the Scala
`timeConf` to plain milliseconds before passing it to native, so both encode
the same 10-minute value and there is a single documented representation
(`10m`).
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
