shuwang21 commented on code in PR #42357:
URL: https://github.com/apache/spark/pull/42357#discussion_r1292889034
##########
resource-managers/yarn/src/main/scala/org/apache/spark/deploy/yarn/config.scala:
##########
@@ -462,6 +462,31 @@ package object config extends Logging {
.stringConf
.createWithDefault("yarn.io/fpga")
+ private[spark] val YARN_CLIENT_STAT_CACHE_PRELOAD_ENABLED =
+ ConfigBuilder("spark.yarn.client.statCache.preload.enabled")
+ .doc("This configuration enables statCache to be preloaded at YARN client
side. This feature" +
+ " analyzes the pattern of resources paths. If multiple resources shared
the same parent" +
+ " directory, a single <code>listStatus</code> will be invoked on the
parent directory" +
+ " instead of multiple <code>getFileStatus</code> performed on each
individual resources." +
+ " If most resources from a small set of directories, this can
substantially improve job" +
+ " submission time. Noticing, this could potentially increase the memory
overhead at client" +
+ " side.")
+ .version("4.0.0")
+ .booleanConf
+ .createWithDefault(false)
+
+ private[spark] val YARN_CLIENT_STAT_CACHE_PRELOAD_PER_DIRECTORY_THRESHOLD =
+ ConfigBuilder("spark.yarn.client.statCache.preload.perDirectoryThreshold")
+ .doc("This configuration defines the threshold for the number of
resources in a directory" +
Review Comment:
Thanks for feedback. Both looks good to me, therefore, i will keep the
current one.
--
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]
