Analytics with Hadoop - Big Data Made Easy: A Working Guide to the Complete Hadoop Toolset - page 257

The permissions string reads d rwx r-x r-x . The d means “directory,” the first three characters are permissions
for the owner, the next set is the group, and the final set is the world (or everybody else). So, the group membership
string has r-x , or read and execute permissions; it is not writeable by the group because the w character is missing.
Also, the string hdfs hdfs indicates that the directory is owned by the user and group hdfs. When I check, I make sure
that the directory is either group writeable by the Linux root user or is not group writeable.
Now, I create the sockets directory and subdirectory under /var/run as the root user:
[root@hc1r1m1 impala]# mkdir -p /var/run/hdfs-sockets/dn
[root@hc1r1m1 impala]# ls -ld /var/run/hdfs-sockets
drwxr-xr-x 3 root root 4096 Sep 7 09:58 /var/run/hdfs-sockets
I start the Impala services (the Impala State Store server, the Catalog server, and the Impala server) as root:
[root@hc1r1m1 ~]# service impala-state-store start
[root@hc1r1m1 ~]# service impala-catalog start
[root@hc1r1m1 ~]# service impala-server start
I check the Impala logs under /var/log/impala for errors. Errors can occur because of incorrect configuration
after installation. For instance, if the sockets directory does not exist, the following error message would appear:
E0907 09:49:42.279753 4815 impala-server.cc:208] ERROR: short-circuit local reads is disabled
because
- Impala cannot read or execute the parent directory of dfs.domain.socket.path
Or, if the configuration value dfs.client.file-block-storage-locations.timeout is either not specified or has too small
a value, the following error message will be issued:
ERROR: block location tracking is not properly enabled because
- dfs.client.file-block-storage-locations.timeout is too low. It should be at least 3000.
E0907 09:49:42.280009 4815 impala-server.cc:210] Aborting Impala Server startup due to improper
configuration
I plan to access Impala via Hue, so on the server hc1nn, I need to change the Hue configuration file hue.ini under
/etc/hue/conf so that Hue knows where to find Impala. To do so, I set the server_host value in the [Impala] section of
the file to the host where Impala is running, hc1r1m1:
server_host=hc1r1m1
After making the change, I restart the Hue server on the host hc1nn as the Linux root user:
[root@hc1nn ~]# service hue restart
The interface is now be ready for use.
Impala User Interfaces
You can access the Impala server user interface via the URL hc1r1m1:25000 . Figure 9-1 shows a basic example of this
interface. The Impala State Store server user interface resides at port number 25010, and is reached using the URL
hc1r1m1:25010. The Impala Catalogue server user interface is on port 25020. All three of these interfaces provide
information such as configuration, log content, metrics, and session. Because I concentrate on Impala's analytics
functionality using SQL, I only examine the Impala server interface and leave you to investigate the other options.