[HIVE-16805] Utilities isEmptyPath Logging Too Chatty and Uses Bad Format

• Type: Improvement
• Status: Closed
• Priority: Minor
• Resolution: Fixed
• Affects Version/s: 2.1.1, 3.0.0
• Fix Version/s: 3.0.0
• Component/s: Query Processor
• Flags: Patch
• Assignee: David Mollitor
• Reporter: David Mollitor
• Created: 01/Jun/17 19:18
• Updated: 22/May/18 23:58
• Resolved: 05/Feb/18 13:55

Description

org.apache.hadoop.hive.ql.exec.Utilities

  public static boolean isEmptyPath(JobConf job, Path dirPath, Context ctx)
      throws Exception {
    if (ctx != null) {
      ContentSummary cs = ctx.getCS(dirPath);
      if (cs != null) {
        LOG.info("Content Summary " + dirPath + "length: " + cs.getLength() + " num files: "
            + cs.getFileCount() + " num directories: " + cs.getDirectoryCount());
        return (cs.getLength() == 0 && cs.getFileCount() == 0 && cs.getDirectoryCount() <= 1);
      } else {
        LOG.info("Content Summary not cached for " + dirPath);
      }
    }
    return isEmptyPath(job, dirPath);
  }

1. This is too chatty. Logging for caching is not necessary for INFO logging. Move to DEBUG
2. Use Log4J's {} format
3. Make the messages more symmetrical - The success message doesn't include the word "cache" in it at all
4. Missing a space between the path and "length: " so the path is concatenated with the string "length:"