Support Questions

Find answers, ask questions, and share your expertise

/user/accumulo permissions problem

avatar
New Contributor

I'm trying to track down an "under replicated blocks" issue and ran into a permission problem. If I run 'fsck' here's what happens:

 

[admin@fatman ~t]$ hdfs fsck /

Connecting to namenode via http://fatman.localdomain:50070

FSCK started by admin (auth:SIMPLE) from /10.1.1.10 for path / at Fri Jan 16 09:24:15 EST 2015

....................................................................................................

....................................................................................................

.............................................FSCK ended at Fri Jan 16 09:24:15 EST 2015 in 42 milliseconds

Permission denied: user=admin, access=READ_EXECUTE, inode="/user/accumulo/.Trash":accumulo:accumulo:drwx------

 

[admin@fatman ~]$ hadoop fs -ls /user/accumulo
Found 1 items
drwx------   - accumulo accumulo          0 2015-01-15 19:00 /user/accumulo/.Trash
[admin@fatman ~]$

 

This indicates that only the 'accumulo' user can access the directory. According to the /etc/passwd file the accumulo account does not support login and therefore I don't know how to become the accumulo user. Sounds like a trap to me!

 

From /etc/passwd:

accumulo:x:477:473:Accumulo:/var/lib/accumulo:/sbin/nologin

 

I believe at this point that I may have chased all the under replicated blocks into the trash folder. The Cloudera Manager idicates that I still have such blocks in HDFS (from the HDFS Health Test):

 

"1,712 under replicated blocks in the cluster. 2,678 total blocks in the cluster. Percentage under replicated blocks: 63.93%. Critical threshold: 40.00%."

 

So...I'm nobody's sys admin...and I don't know where to go from here. Can anyone give me a hint about how to resolve this permissions issue?  A secondary question, is there a way I can empty the accumulo trash folder short of executing a file system command that would involve the aforementioned permission problem?

 

 

Environment: CDH5.3 with Accumulo installed using Cloudera Manager. Small cluster with 4 data nodes.

1 ACCEPTED SOLUTION

avatar
New Contributor

I figured out how to get around the permissions problem. Simply have to 'su' to the hdfs user, which is the super user for HDFS

 

[root@fatman ~]# su - hdfs

-bash-4.1$ hdfs fsck /

 

That works.

 

View solution in original post

1 REPLY 1

avatar
New Contributor

I figured out how to get around the permissions problem. Simply have to 'su' to the hdfs user, which is the super user for HDFS

 

[root@fatman ~]# su - hdfs

-bash-4.1$ hdfs fsck /

 

That works.