question Re: Duplicate Directories in HDFS in Archives of Support Questions (Read Only)

Duplicate Directories in HDFS

Srini — Fri, 16 Sep 2022 10:03:36 GMT

Hi All,

Our application team created hdfs directories with below script.

hadoop fs -mkdir /a/b/c/d/20160208
hadoop fs -mkdir /a/b/c/d/20160208/s
hadoop fs -mkdir /a/b/c/d/20160208/s/inputmap
hadoop fs -mkdir /a/b/c/d/20160208/s/temp
hadoop fs -mkdir /a/b/c/d/20160208/s/map
hadoop fs -mkdir /a/b/c/d/20160208/s/input
hadoop fs -copyFromLocal /x/y/z/20160208.dat /a/b/c/d/20160208/s/inputmap

echo "Setup Complete"

The directories got creted but it throws error if we try to access it.

hdfs@hostname$ hadoop fs -ls /a/b/c/d/
Found 20 items
drwxr-xr-x - user group 0 2016-01-27 09:10 /a/b/c/d/20141211
drwxr-xr-x - user group 0 2016-01-06 01:03 /a/b/c/d/20141212
drwxr-xr-x - user group 0 2016-01-06 01:09 /a/b/c/d/20141213
drwxr-xr-x - user group 0 2015-11-12 08:53 /a/b/c/d/20151106
drwxr-xr-x - user group 0 2016-01-12 01:48 /a/b/c/d/20151118
drwxr-xr-x - user group 0 2015-12-04 04:21 /a/b/c/d/20151130
drwxrwxr-x - user group 0 2016-01-12 10:48 /a/b/c/d/20151221
drwxr-xr-x - user group 0 2016-01-19 11:23 /a/b/c/d/20160111
drwxr-xr-x - user group 0 2016-01-27 14:56 /a/b/c/d/20160112
drwxr-xr-x - user group 0 2016-02-02 16:12 /a/b/c/d/20160125
drwxr-xr-x - user group 0 2016-02-08 12:41 /a/b/c/d/20160126
drwxr-xr-x - user group 0 2016-02-08 10:26 /a/b/c/d/20160127
drwxr-xr-x - user group 0 2016-01-29 10:48 /a/b/c/d/20160129
drwxr-xr-x - user group 0 2016-02-09 02:43 /a/b/c/d/20160203
drwxr-xr-x - user group 0 2016-02-09 02:42 /a/b/c/d/20160204
drwxr-xr-x - user group 0 2016-02-08 15:38 /a/b/c/d/20160205
drwxr-xr-x - user group 0 2016-02-08 09:02 /a/b/c/d/20160205
drwxr-xr-x - user group 0 2016-02-08 07:00 /a/b/c/d/20160206
drwxr-xr-x - user group 0 2016-02-09 17:11 /a/b/c/d/20160208
drwxr-xr-x - user group 0 2016-02-08 11:07 /a/b/c/d/20160208
hdfs@hostname$ hadoop fs -ls /a/b/c/d/20160206
ls: `/a/b/c/d/20160206': No such file or directory

when we did | cat -v along with "ls" we came to know that some special character got inserted in the directory name as below.

hdfs@hostname$ hadoop fs -ls /a/b/c/d/ | cat -v
Found 20 items
drwxr-xr-x - user group 0 2016-01-27 09:10 /a/b/c/d//20141211
drwxr-xr-x - user group 0 2016-01-06 01:03 /a/b/c/d//20141212
drwxr-xr-x - user group 0 2016-01-06 01:09 /a/b/c/d//20141213
drwxr-xr-x - user group 0 2015-11-12 08:53 /a/b/c/d//20151106
drwxr-xr-x - user group 0 2016-01-12 01:48 /a/b/c/d//20151118
drwxr-xr-x - user group 0 2015-12-04 04:21 /a/b/c/d//20151130
drwxrwxr-x - user group 0 2016-01-12 10:48 /a/b/c/d//20151221
drwxr-xr-x - user group 0 2016-01-19 11:23 /a/b/c/d//20160111
drwxr-xr-x - user group 0 2016-01-27 14:56 /a/b/c/d//20160112
drwxr-xr-x - user group 0 2016-02-02 16:12 /a/b/c/d//20160125
drwxr-xr-x - user group 0 2016-02-08 12:41 /a/b/c/d//20160126
drwxr-xr-x - user group 0 2016-02-08 10:26 /a/b/c/d//20160127
drwxr-xr-x - user group 0 2016-01-29 10:48 /a/b/c/d//20160129
drwxr-xr-x - user group 0 2016-02-09 02:43 /a/b/c/d//20160203
drwxr-xr-x - user group 0 2016-02-09 02:42 /a/b/c/d//20160204
drwxr-xr-x - user group 0 2016-02-08 15:38 /a/b/c/d//20160205
drwxr-xr-x - user group 0 2016-02-08 09:02 /a/b/c/d//20160205^M
drwxr-xr-x - user group 0 2016-02-08 07:00 /a/b/c/d//20160206^M
drwxr-xr-x - user group 0 2016-02-09 17:11 /a/b/c/d//20160208
drwxr-xr-x - user group 0 2016-02-08 11:07 /a/b/c/d//20160208^M

Now i want to delete these duplicate entries, can anyone help me with this.

Thanks

Srini

Re: Duplicate Directories in HDFS

denloe — Thu, 11 Feb 2016 16:40:34 GMT

You would handle this in the same way if the issue occurred on a linux filesystem. Use quotes around the filename and ctrl-v to insert the special characters.

In this case, I type ctrl-v then ctrl-m to insert ^M into my strings.

$ hdfs dfs -put /etc/group "/tmp/abc^M"

$ hdfs dfs -ls /tmp
Found 4 items
drwxrwxrwx   - hdfs   supergroup          0 2016-02-11 11:29 /tmp/.cloudera_health_monitoring_canary_files
-rw-r--r--   3 hdfs   supergroup        954 2016-02-11 11:30 /tmp/abc
drwx-wx-wx   - hive   supergroup          0 2016-01-11 12:10 /tmp/hive
drwxrwxrwt   - mapred hadoop              0 2016-01-11 12:08 /tmp/logs

$ hdfs dfs -ls /tmp | cat -v
Found 4 items
drwxrwxrwx   - hdfs   supergroup          0 2016-02-11 11:30 /tmp/.cloudera_health_monitoring_canary_files
-rw-r--r--   3 hdfs   supergroup        954 2016-02-11 11:30 /tmp/abc^M
drwx-wx-wx   - hive   supergroup          0 2016-01-11 12:10 /tmp/hive
drwxrwxrwt   - mapred hadoop              0 2016-01-11 12:08 /tmp/logs

$ hdfs dfs -mv "/tmp/abc^M" /tmp/abc

$ hdfs dfs -ls /tmp | cat -v
Found 4 items
drwxrwxrwx   - hdfs   supergroup          0 2016-02-11 11:31 /tmp/.cloudera_health_monitoring_canary_files
-rw-r--r--   3 hdfs   supergroup        954 2016-02-11 11:30 /tmp/abc
drwx-wx-wx   - hive   supergroup          0 2016-01-11 12:10 /tmp/hive
drwxrwxrwt   - mapred hadoop              0 2016-01-11 12:08 /tmp/logs

Re: Duplicate Directories in HDFS

denloe — Thu, 11 Feb 2016 16:48:51 GMT

In my example I used -mv. You would use -rmdir.

hdfs dfs -rmdir "/a/b/c/d//20160205^M"

Remember, to get "^M" type ctrl-v ctrl-m.

Re: Duplicate Directories in HDFS

sathishkumar — Thu, 11 Feb 2016 17:23:48 GMT

Thanks Denloe for your response. I actually got one more doubt.

What if there is some other name following with after that ^M ? Should we need to use

hdfs dfs -rmdir "/a/b/c/d//20160205^Msomepart" or do we need to use some escape sequence for this?

Moreover when I press ctrl+v or ctrl+m the command is immediately executing (it is not allowing to type the second word)

Re: Duplicate Directories in HDFS

denloe — Thu, 11 Feb 2016 19:21:11 GMT

The non-printable character may be located anywhere in the filename. You just need to insert it in the appropriate location when quoting the filename.

Using ctrl-v to insert special characters is the default for the bash shell, but your terminal emulator (especially if you are coming in from Windows) may be catching it instead.

Try using shift-insert instead of ctrl-v. If that fails, you may need to find an alternate method to embed control characters, such as use vi to create to bash script and insert them using vi.

Re: Duplicate Directories in HDFS

Harsh J — Fri, 12 Feb 2016 05:29:14 GMT

I prefer using the simpler bash syntax of using special escaped characters,
if it helps:

We know that ^M is the same as \r, which makes sense if you used Windows
Notepad to write the commands but forgot to convert the file via dos2unix:

~> echo $'\x0d' | cat -v
^M
~> echo -n $'\x0d' | od -c
0000000 \r
0000002

(The \x0D or \x0d is the hex equivalent of \r, per
http://www.asciitable.com/ (carriage return))

Therefore, you can use the $'' syntax to write a string that includes the
escape:

~> hadoop fs -ls $'/a/b/c/d/20160206\r'
Or,
~> hadoop fs -ls $'/a/b/c/d/20160206\x0d'

This words well regardless of the terminal emulator you are using, cause
we're escaping based on representation vs. by reliance on the emulator
understanding the characters via input.

Re: Duplicate Directories in HDFS

ianeeshps — Tue, 24 Oct 2017 07:12:55 GMT

There is a simple method to remove those.

1. List those directories inside a txt file like below

hadoop fs -ls /path > test

2. cat -t test will give you positions of duplicate with junk character

3. open another shell and just try to comment it # to identify exact ones

4. again cat -t the file to confirm u commented the culprits

5. remove original folder frm list

6. for i in `cat list`;

do hadoop fs -rmr $i;

done