External PostgreSQL Database
To use an external PostgreSQL database, follow these procedures.
Installing the External PostgreSQL Server
- If you already have a PostgreSQL database set up, you can skip to the section Configuring and Starting the PostgreSQL Server to verify that your PostgreSQL configurations meet the requirements for Cloudera Manager.
- Make sure that the data directory, which by default is /var/lib/postgresql/data/, is on a partition that has sufficient free space.
- Use one or more of the following commands to set the locale:
export LANGUAGE=en_US.UTF-8 export LANG=en_US.UTF-8 export LC_ALL=en_US.UTF-8 locale-gen en_US.UTF-8 dpkg-reconfigure locales
- Install PostgreSQL packages:
- Red Hat
$ sudo yum install postgresql-server
$ sudo zypper install postgresql91-serverNote
:This command will install PostgreSQL 9.1. If you want to install a different version, you can use zypper search postgresql to search for available versions. You should install version 8.4 or higher.
$ sudo apt-get install postgresql
- Red Hat
Configuring and Starting the PostgreSQL Server
By default, PostgreSQL only accepts connections on the loopback interface. You must reconfigure PostgreSQL to accept connections from the Fully Qualified Domain Name (FQDN) of the hosts hosting the management roles. If you do not make these changes, the management processes cannot connect to and use the database on which they depend.
- Initialize the external PostgreSQL database. For some versions of
PostgreSQL, this occurs automatically the first time that you start the
PostgreSQL server. In this case, issue the
$ sudo service postgresql startIn other versions, you must explicitly initialize the database using:
$ sudo service postgresql initdbSee the PostgreSQL documentation for more details.
- Enable MD5 authentication. Edit pg_hba.conf, which is usually found in /var/lib/pgsql/data or /etc/postgresql/8.4/main. Add the
host all all 127.0.0.1/32 md5If the default pg_hba.conf file contains the following line:
host all all 127.0.0.1/32 identthen the host line specifying md5 authentication shown above must be inserted before this ident line. Failure to do so may cause an authentication error when running the scm_prepare_database.sh script. You can modify the contents of the md5 line shown above to support different configurations. For example, if you want to access PostgreSQL from a different host, replace 127.0.0.1 with your IP address and update postgresql.conf, which is typically found in the same place as pg_hba.conf, to include:
listen_addresses = '*'
- Configure settings to ensure your system performs as expected. Update
these settings in the /var/lib/pgsql/data/postgresql.conf or /var/lib/postgresql/data/postgresql.conf file. Settings vary based
on cluster size and resources as follows:
- Small to mid-sized clusters - Consider the following
settings as starting points. If resources are limited, consider reducing
the buffer sizes and checkpoint segments further. Ongoing tuning may be
required based on each host's resource utilization. For example, if the
Cloudera Manager Server is running on the same host as other roles, the
following values may be acceptable:
- shared_buffers - 256MB
- wal_buffers - 8MB
- checkpoint_segments - 16
- checkpoint_completion_target - 0.9
- Large clusters - Can contain up to 1000 hosts.
Consider the following settings as starting points.
- max_connection - For large clusters, each database is typically hosted on a different host. In general, allow each database on a host 100 maximum connections and then add 50 extra connections. You may have to increase the system resources available to PostgreSQL, as described at Connection Settings.
- shared_buffers - 1024 MB. This requires that the operating system can allocate sufficient shared memory. See PostgreSQL information on Managing Kernel Resources for more information on setting kernel resources.
- wal_buffers - 16 MB. This value is derived from the shared_buffers value. Setting wal_buffers to be approximately 3% of shared_buffers up to a maximum of approximately 16 MB is sufficient in most cases.
- checkpoint_segments - 128. The PostgreSQL Tuning Guide recommends values between 32 and 256 for write-intensive systems, such as this one.
- checkpoint_completion_target - 0.9. This setting is only available in PostgreSQL versions 8.3 and later, which are highly recommended.
- Small to mid-sized clusters - Consider the following settings as starting points. If resources are limited, consider reducing the buffer sizes and checkpoint segments further. Ongoing tuning may be required based on each host's resource utilization. For example, if the Cloudera Manager Server is running on the same host as other roles, the following values may be acceptable:
- Configure the PostgreSQL server to start at boot.
- Red Hat
$ sudo /sbin/chkconfig postgresql on $ sudo /sbin/chkconfig --list postgresql postgresql 0:off 1:off 2:on 3:on 4:on 5:on 6:off
$ sudo chkconfig --add postgresql
$ sudo chkconfig postgresql on
- Red Hat
- Start or restart the PostgreSQL database:
$ sudo service postgresql restart
Creating Databases for Activity Monitor, Reports Manager, Hive Metastore, Sentry Server, Cloudera Navigator Audit Server, and Cloudera Navigator Metadata Server
- If you are not using the Cloudera Manager installer, the Cloudera Manager Server.
- Cloudera Management Service roles:
- Activity Monitor (if using the MapReduce service)
- Reports Manager
- Each Hive metastore
- Sentry Server
- Cloudera Navigator Audit Server
- Cloudera Navigator Metadata Server
You can create these databases on the host where the Cloudera Manager Server will run, or on any other hosts in the cluster. For performance reasons, you should install each database on the host on which the service runs, as determined by the roles you assign during installation or upgrade. In larger deployments or in cases where database administrators are managing the databases the services use, you can separate databases from services, but use caution.
The database must be configured to support UTF-8 character set encoding.
Record the values you enter for database names, user names, and passwords. The Cloudera Manager installation wizard requires this information to correctly connect to these databases.
- Connect to PostgreSQL:
$ sudo -u postgres psql
- If you are not using the Cloudera Manager installer, create a database
for the Cloudera Manager Server. The database name, user name, and password
can be any value. Record the names chosen because you will need them later
when running the scm_prepare_database.sh
postgres=# CREATE ROLE scm LOGIN PASSWORD 'scm'; postgres=# CREATE DATABASE scm OWNER scm ENCODING 'UTF8';
- Create databases for Activity Monitor, Reports Manager, Hive Metastore, Sentry
Server, Cloudera Navigator Audit Server, and Cloudera Navigator Metadata
postgres=# CREATE ROLE user LOGIN PASSWORD 'password'; postgres=# CREATE DATABASE databaseName OWNER user ENCODING 'UTF8';where user, password, and databaseName can be any value. The examples shown match the default names provided in the Cloudera Manager configuration settings:For PostgreSQL 8.2.23 or later, also run:
Role Database User Password Activity Monitor amon amon amon_password Reports Manager rman rman rman_password Hive Metastore Server metastore hive hive_password Sentry Server sentry sentry sentry_password Cloudera Navigator Audit Server nav nav nav_password Cloudera Navigator Metadata Server navms navms navms_password
postgres=# ALTER DATABASE Metastore SET standard_conforming_strings = off;