diff --git a/core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala b/core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala
index 1311cf8607d1c..56135d854453b 100644
--- a/core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala
+++ b/core/src/main/scala/org/apache/spark/scheduler/EventLoggingListener.scala
@@ -21,8 +21,9 @@ import java.io._
 import java.net.URI
 import java.nio.charset.StandardCharsets
 import java.util.Locale
+import java.util.concurrent.ConcurrentHashMap
+import java.util.function.Function
 
-import scala.collection.mutable
 import scala.collection.mutable.ArrayBuffer
 
 import org.apache.hadoop.conf.Configuration
@@ -290,7 +291,7 @@ private[spark] object EventLoggingListener extends Logging {
   private val LOG_FILE_PERMISSIONS = new FsPermission(Integer.parseInt("770", 8).toShort)
 
   // A cache for compression codecs to avoid creating the same codec many times
-  private val codecMap = new mutable.HashMap[String, CompressionCodec]
+  private val codecMap = new ConcurrentHashMap[String, CompressionCodec]
 
   /**
    * Write metadata about an event log to the given stream.
@@ -357,7 +358,10 @@ private[spark] object EventLoggingListener extends Logging {
     val in = new BufferedInputStream(fs.open(log))
     try {
       val codec = codecName(log).map { c =>
-        codecMap.getOrElseUpdate(c, CompressionCodec.createCodec(new SparkConf, c))
+        codecMap.computeIfAbsent(c, new Function[String, CompressionCodec] {
+          override def apply(key: String): CompressionCodec =
+            CompressionCodec.createCodec(new SparkConf, key)
+        })
       }
       codec.map(_.compressedInputStream(in)).getOrElse(in)
     } catch {