CDH 5 Packaging and Tarball Information

Each CDH release series is made up of a collection of CDH project packages that are known to work together. The package version numbers of the CDH projects in each CDH release are listed in the following table.

Examples of Versions

Cloudera packages are designed to be transparent and easy to customize. CDH 5 packages are labeled using the following format:

component-base_version+cdh_version+patch_level

where,

  • component-base_version is the version of the open-source component included in the CDH package
  • cdh_version is the version of the CDH package
  • patch_level is the number of source commits applied on top of the base version forked from the Apache Hadoop branch. The list of source commits includes all backports and all non-functional changes such as CDH packaging and branding commits. Note that the number of commits does not indicate the number of functional changes or bug fixes in the release. For example, a commit may be used to amend a version number or make other non-functional changes. The list of actual patches that will match the count is found inside the CDH tarball's ../cloudera/patches/ directory. All of the source commits Cloudera has applied is also available in source form in that directory.
For example:

Package

Component

Branch

Base Version

CDH Version

Patch Level

hadoop-2.3.0+cdh5.1.0+795

hadoop

2.3

2.3.0

5.1.0

795

hue-3.6.0+cdh5.1.0+86

hue

3.6 

3.6.0

5.1.0

86

parquet-1.2.5+cdh5.1.0+130

parquet

1.2

1.2.5

5.1.0

130

CDH Package Manifests

Both the CDH patched source and packages contain explicit information about Cloudera modifications. For example, in the patched source there is a top-level cloudera directory with:

  • A CHANGES.cloudera.txt file that lists all the changes to the pristine source
  • A patches directory that contains every patch Cloudera has applied to the pristine source. All Cloudera patches are released with an Apache 2.0 license.
  • A files directory for files Cloudera created from scratch, such as man pages and configuration files. All Cloudera files are released with an Apache 2.0 license.
  • A README.cloudera file that explains explicitly how to recreate the patches source from the pristine source.

Build and Release Numbering

If you are installing CDH 5 with a package manager, you will also see build and release information as part of the file name. The build and package release fields follow the patch level: for example, hbase-0.98.1+cdh5.1.0+64-1.cdh5.1.0.p0.34.el6.x86_64.rpm. The suffix -1.cdh5.1.0.p0.34.el6.x86_64 represents:

  • the base of the release field (1)
  • the CDH release (cdh5.1.0)
  • the customer patch identifier (p0 — which will be 0 for all regular CDH releases but will increment for customer patches)
  • the build number (34)
  • the distribution (el6 = RHEL/CentOs 6, el5 = RHEL/Centos 5, sles11 = SLES 11)
  • the processor architecture (x86_64, noarch, i386, amd_64). noarch means the packages are not architecture-specific.

External Documentation

Cloudera provides documentation for CDH as a whole, whether your CDH cluster is managed by Cloudera Manager or not. In addition, you may find it useful to refer to documentation for the individual components included in CDH. Where possible, these links point to the main documentation for a project, in the Cloudera release archive. This ensures that you are looking at the correct documentation for the version of a project included in CDH. Otherwise, the links may point to the project's main site.