Skip to content

Lazy metrics initialization (to correct pick up in executor) #1195

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
feast-ci-bot merged 1 commit into from
Nov 24, 2020
Merged

Lazy metrics initialization (to correct pick up in executor) #1195

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 2 additions & 6 deletions spark/ingestion/src/main/scala/feast/ingestion/BasePipeline.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,11 +41,7 @@ trait BasePipeline {
case Some(c: StatsDConfig) =>
conf
.set(
"spark.metrics.conf.*.source.redis.class",
"org.apache.spark.metrics.source.RedisSinkMetricSource"
)
.set(
"spark.metrics.conf.*.source.redis.labels",
"spark.metrics.labels",
s"feature_table=${jobConfig.featureTable.name}"
)
.set(
Expand All @@ -56,7 +52,7 @@ trait BasePipeline {
.set("spark.metrics.conf.*.sink.statsd.port", c.port.toString)
.set("spark.metrics.conf.*.sink.statsd.period", "30")
.set("spark.metrics.conf.*.sink.statsd.unit", "seconds")
.set("spark.metrics.namespace", jobConfig.mode.toString)
.set("spark.metrics.namespace", jobConfig.mode.toString.toLowerCase)
case None => ()
}

Expand Down
17 changes: 13 additions & 4 deletions spark/ingestion/src/main/scala/feast/ingestion/metrics/StatsdReporterWithTags.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -169,19 +169,28 @@ class StatsdReporterWithTags(
reportMetered(name, timer)
}

private val nameWithTag = """(\S+)#(\S+)""".r

private def send(name: String, value: String, metricType: String)(implicit
socket: DatagramSocket
): Unit = {
val bytes = sanitize(s"$name:$value|$metricType").getBytes(UTF_8)
val bytes = name match {
case nameWithTag(name, tags) =>
val tagsWithSemicolon = tags.replace('=', ':')
sanitize(s"$name:$value|$metricType|#$tagsWithSemicolon").getBytes(UTF_8)
case _ =>
sanitize(s"$name:$value|$metricType").getBytes(UTF_8)
}
val packet = new DatagramPacket(bytes, bytes.length, address)
socket.send(packet)
}

private val nameWithTag = """(\S+)#(\S+)""".r

private def fullName(name: String, suffixes: String*): String = name match {
case nameWithTag(name, tags) =>
MetricRegistry.name(prefix, name +: suffixes: _*) ++ "#" ++ tags
// filter out parts that consists only from numbers
// that could be executor-id for example
val stableName = name.split('.').filterNot(_ forall Character.isDigit).mkString(".")
MetricRegistry.name(prefix, stableName +: suffixes: _*) ++ "#" ++ tags
case _ =>
MetricRegistry.name(prefix, name +: suffixes: _*)
}
Expand Down
19 changes: 18 additions & 1 deletion spark/ingestion/src/main/scala/feast/ingestion/stores/redis/RedisSinkRelation.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,9 @@ class RedisSinkRelation(override val sqlContext: SQLContext, config: SparkRedisC
extends BaseRelation
with InsertableRelation
with Serializable {

import RedisSinkRelation._

private implicit val redisConfig: RedisConfig = {
new RedisConfig(
new RedisEndpoint(sqlContext.sparkContext.getConf)
Expand Down Expand Up @@ -151,11 +154,21 @@ class RedisSinkRelation(override val sqlContext: SQLContext, config: SparkRedisC
.build
}

private lazy val metricSource: Option[RedisSinkMetricSource] =
private lazy val metricSource: Option[RedisSinkMetricSource] = {
MetricInitializationLock.synchronized {
// RedisSinkMetricSource needs to be registered on executor and SparkEnv must already exist.
// Which is problematic, since metrics system is initialized before SparkEnv set.
// That's why I moved source registering here
if (SparkEnv.get.metricsSystem.getSourcesByName(RedisSinkMetricSource.sourceName).isEmpty) {
SparkEnv.get.metricsSystem.registerSource(new RedisSinkMetricSource)
}
}

SparkEnv.get.metricsSystem.getSourcesByName(RedisSinkMetricSource.sourceName) match {
case Seq(head) => Some(head.asInstanceOf[RedisSinkMetricSource])
case _ => None
}
}

private def groupKeysByNode(
nodes: Array[RedisNode],
Expand Down Expand Up @@ -202,3 +215,7 @@ class RedisSinkRelation(override val sqlContext: SQLContext, config: SparkRedisC

}
}

object RedisSinkRelation {
object MetricInitializationLock
}
15 changes: 10 additions & 5 deletions spark/ingestion/src/main/scala/org/apache/spark/metrics/source/RedisSinkMetricSource.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,23 +17,28 @@
package org.apache.spark.metrics.source

import com.codahale.metrics.MetricRegistry
import org.apache.spark.{SparkConf, SparkEnv}
import org.apache.spark.SparkEnv

class RedisSinkMetricSource extends Source {
override val sourceName: String = RedisSinkMetricSource.sourceName

override val metricRegistry: MetricRegistry = new MetricRegistry

private val sparkConfig = Option(SparkEnv.get).map(_.conf).getOrElse(new SparkConf(true))
private val sparkConfig = SparkEnv.get.conf

private val metricLabels = sparkConfig.get("spark.metrics.conf.*.source.redis.labels")
private val metricLabels = sparkConfig.get("spark.metrics.labels", "")

private def nameWithLabels(name: String) =
private val appId = sparkConfig.get("spark.app.id", "")

private val executorId = sparkConfig.get("spark.executor.id", "")

private def nameWithLabels(name: String) = {
if (metricLabels.isEmpty) {
name
} else {
s"$name#$metricLabels"
s"$name#$metricLabels,job_id=$appId-$executorId"
}
}

val METRIC_TOTAL_ROWS_INSERTED =
metricRegistry.counter(nameWithLabels("feast_ingestion_feature_row_ingested_count"))
Expand Down
6 changes: 3 additions & 3 deletions spark/ingestion/src/test/scala/feast/ingestion/metrics/StatsReporterSpec.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,19 +89,19 @@ class StatsReporterSpec extends UnitSpec {
server.receive should contain("test:0|g")
}

"Statsd reporter" should "keep tags part in the name's end" in new Scope {
"Statsd reporter" should "keep tags part in the message's end" in new Scope {
reporter.report(
gauges = Collections.emptySortedMap(),
counters = Collections.emptySortedMap(),
histograms = new util.TreeMap(
Map(
"test#fs=name" -> histogram((1 to 100))
"prefix.1111.test#fs=name,job=aaa" -> histogram((1 to 100))
).asJava
),
meters = Collections.emptySortedMap(),
timers = Collections.emptySortedMap()
)

server.receive should contain("test.p95#fs=name:95.95|ms")
server.receive should contain("prefix.test.p95:95.95|ms|#fs:name,job:aaa")
}
}