[HADOOP-714] Namenode format does not work when hadoop is installed for first time

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Duplicate
• Affects Version/s: 0.8.0
• Fix Version/s: 0.9.0

Environment

Linux CentOS 4

• Assignee: Unassigned
• Reporter: Lei Chen
• Created: 14/Nov/06 00:56
• Updated: 08/Jul/09 16:42
• Resolved: 14/Nov/06 18:19

Description

The namenode can not be formatted for the first time installation. There is a bug is the main function of org.apache.hadoop.dfs.NameNode class. The code will skip formatting when the directory for name does not exist. There is a possible fix for the bug

if (argv.length == 1 && argv[0].equals("-format")) {
boolean aborted = false;
File[] dirs = getDirs(conf);

for (int idx = 0; idx < dirs.length; idx++) {
if (dirs[idx].exists()) {
System.err.print("Re-format filesystem in " + dirs[idx]
+ " ? (Y or N) ");
if (!(System.in.read() == 'Y'))

{ System.err .println("Format aborted in " + dirs[idx]); aborted = true; }

else

{ System.out.println("do format" + dirs[idx].getAbsolutePath()); format(dirs[idx]); System.err.println("Formatted " + dirs[idx]); }

System.in.read(); // discard the enter-key
} else

{ System.out.println("do format" + dirs[idx].getAbsolutePath()); format(dirs[idx]); System.err.println("Formatted " + dirs[idx]); }

}

System.exit(aborted ? 1 : 0);
}