steveloughran commented on pull request #2069:
URL: https://github.com/apache/hadoop/pull/2069#issuecomment-661077889


   The latest update records the min/max and mean times to initiate the 
(long-haul) http get request
   
   ```
   2020-07-17 17:56:30,433 [JUnit] INFO  scale.ITestS3AInputStreamPerformance 
(ITestS3AInputStreamPerformance.java:dumpIOStatistics(135)) -
   Aggregate Stream Statistics counters=((stream_aborted=2) 
(stream_read_bytes=47870126) (stream_read_bytes_backwards_on_seek=12713984) 
   (stream_read_bytes_discarded_in_abort=43889622) 
(stream_read_bytes_read_in_close=252395) 
(stream_read_bytes_skipped_on_seek=55054163)
   (stream_read_close_operations=0) (stream_read_closed=12) 
(stream_read_exceptions=0) (stream_read_fully_operations=8)
   (stream_read_opened=14) (stream_read_operations=3415) 
(stream_read_operations_incomplete=3362)
   (stream_read_seek_backward_operations=4) 
(stream_read_seek_bytes_read=45092691) 
   (stream_read_seek_forward_operations=175) (stream_read_seek_operations=179) 
   (stream_read_seek_policy_changed=8) (stream_read_total_bytes=93215212) 
   (stream_read_version_mismatches=0)); 
gauges=((stream_read_gauge_input_policy=6)); 
   minimums=((op_http_get_request.min=29));
   maximums=((op_http_get_request.max=753));
   means=((op_http_get_request.mean=MeanStatistic{sum=2420, samples=14, 
mean=172.85714285714286})); 
   ```
   
   S3A also collects it for listings, and pass that all the way back through 
LocationStatusFetcher
   
   ```
   2020-07-20 14:48:40,563 [JUnit-testLocatedFileStatusFourThreads[raw]] INFO  
s3a.ITestLocatedFileStatusFetcher 
(ITestLocatedFileStatusFetcher.java:assertListCount(184))
   - Statistics of fetcher: counters=((op_http_list_request=4)); gauges=(); 
   minimums=((op_http_list_request.min=29));
   maximums=((op_http_list_request.max=114));
   means=((op_http_list_request.mean=sum=274, samples=4, mean=68.500000)); 
   ```
   
   also goes through LineReader and the codec in/out streams.
   
   This means that applications using the MR classes can now ask for FS 
performance values.
   
   1. o.a.h.fs.statistics API is ready for review; this is the bit we need to 
keep stable.
   2. o.a.h.fs.statistics.impl is also up for looking at. This is where we can 
be agile about change, but we should still look for obvious issues.
   3. S3A stats migration is complete. It's a big part of this patch, but can 
be reviewed independently. It's just driven the work, especially the .impl 
package
   


----------------------------------------------------------------
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.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: common-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: common-issues-h...@hadoop.apache.org

Reply via email to