GeorgeJahad opened a new pull request, #5787: URL: https://github.com/apache/ozone/pull/5787
## What changes were proposed in this pull request? Improving hdfs perf by eliminating redundant config reads, and replacing them with a single one when the fs is initialized. ## What is the link to the Apache JIRA https://issues.apache.org/jira/browse/HDDS-9910 ## How was this patch tested? Current tests still apply. Rename perf under spark improved by a factor of 5: ``` spark-shell --master local --conf spark.kerberos.access.hadoopFileSystems=ofs://vol1 --conf spark.ui.enabled=false --conf spark.port.maxRetries=64 // The following commands are all run from within the sparkshell // init the fs import org.apache.hadoop.fs.Path val ozone_fs = org.apache.hadoop.fs.FileSystem.get(new java.net.URI("ofs://vol1/bdpe/squirrelfso"), sc.hadoopConfiguration) ozone_fs.delete(new Path(s"ofs://vol1/bdpe/squirrelfso/file1/part10001")) ozone_fs.copyFromLocalFile(new Path(s"/etc/hosts"), new Path(s"ofs://vol1/bdpe/squirrelfso/file1/part1")) // run the test spark.time(Range.inclusive(1, 10000).foreach(i => ozone_fs.rename(new Path(s"ofs://vol1/bdpe/squirrelfso/file1/part${i.toString()}"), new Path(s"ofs://vol1/bdpe/squirrelfso/file1/part${(i + 1).toString()}")))) Time taken: 17763 ms ``` -- 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]
