[HDFS-3652] 1.x: FSEditLog failure removes the wrong edit stream when storage dirs have same name

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.0.3, 1.1.0
• Fix Version/s: 1.0.4
• Component/s: namenode
• Target Version/s: 1.0.4
• Hadoop Flags: Reviewed
• Assignee: Todd Lipcon
• Reporter: Todd Lipcon
• Created: 12/Jul/12 23:16
• Updated: 17/Oct/12 18:25
• Resolved: 13/Jul/12 05:01

Description

In FSEditLog.removeEditsForStorageDir, we iterate over the edits streams trying to find the stream corresponding to a given dir. To check equality, we currently use the following condition:

      File parentDir = getStorageDirForStream(idx);
      if (parentDir.getName().equals(sd.getRoot().getName())) {

... which is horribly incorrect. If two or more storage dirs happen to have the same terminal path component (eg /data/1/nn and /data/2/nn) then it will pick the wrong stream(s) to remove.