[HADOOP-9817] FileSystem#globStatus and FileContext#globStatus need to work with symlinks

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 2.3.0
• Fix Version/s: 2.3.0
• Assignee: Colin McCabe
• Reporter: Colin McCabe
• Created: 31/Jul/13 21:46
• Updated: 30/Jun/15 07:22
• Resolved: 05/Aug/13 23:46

Description

FileSystem#globStatus and FileContext#globStatus need to work with symlinks. Currently, they resolve all links, so that if you have:

/alpha/beta
/alphaLink -> alpha

and you take globStatus(/alphaLink/*), you will get /alpha/beta, rather than the expected /alphaLink/beta.

We even resolve terminal symlinks, which would prevent listing a symlink in FSShell, for example. Instead, we should build up the path incrementally. This will allow the shell to behave as expected, and also allow custom globbers to "see" the correct paths for symlinks.

Issue Links

• blocks: HDFS-4019 FSShell should support creating symlinks
• breaks: HADOOP-9877 Fix listing of snapshot directories in globStatus
• breaks: HADOOP-9887 globStatus does not correctly handle paths starting with a drive spec on Windows
• breaks: HDFS-5093 TestGlobPaths should re-use the MiniDFSCluster to avoid failure on Windows
• is duplicated by: HADOOP-7344 globStatus doesn't grok groupings with a slash
• is duplicated by: HDFS-5021 FSShell delete and FSShell rename should operate on symlinks, not their targets
• is related to: HADOOP-9865 FileContext.globStatus() has a regression with respect to relative path
• is related to: HADOOP-9847 TestGlobPath symlink tests fail to cleanup properly