This page is prepared for Hadoop Core committers. You need committer rights to create a new Hadoop Core release.
These instructions have been updated for Hadoop 0.20.x and 1.x releases. For earlier releases, check out an older revision of this page. For 0.21.0 and later many of the steps need to be done in turn for Common, HDFS, and MapReduce. For 0.23.x and 2.x releases, there is a new version of this page at HowToReleasePostMavenization.
Preparation
- Bulk update Jira to unassign from this release all issues that are open non-blockers and send follow-up notification to the developer list that this was done.
- If you have not already done so, update your @apache.org account via id.apache.org with your key; also add and commit your public key to the Hadoop repository KEYS, appending the output of the following commands:
and publish your key at http://pgp.mit.edu/. Once you commit your changes, log into
gpg --armor --fingerprint --list-sigs <keyid> gpg --armor --export <keyid>
people.apache.org
and pull updates to/www/www.apache.org/dist/hadoop/core
. For more details on signing releases, see Signing Releases and Step-By-Step Guide to Mirroring Releases. - To deploy artifacts to the Apache Maven repository create
~/.m2/settings.xml
:<settings xmlns="http://maven.apache.org/SETTINGS/1.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/SETTINGS/1.0.0 http://maven.apache.org/xsd/settings-1.0.0.xsd"> <servers> <server> <id>apache.staging.https</id> <username>Apache username</username> <password>Apache password</password> </server> </servers> </settings>
Branching
Skip this section if this is NOT the first release in a series (i.e. release X.Y.0).
- Notify developers on the #hadoop IRC channel that you are about to branch a release.
- Update
CHANGES.txt
to include the release version and date (useUnreleased
for the date if it is unknown) and removeTrunk (unreleased changes)
. - Commit these changes to trunk.
svn commit -m "Preparing for release X.Y.Z"
- Create a branch for the release series:
svn copy https://svn.apache.org/repos/asf/hadoop/common/trunk \ https://svn.apache.org/repos/asf/hadoop/common/branches/branch-X.Y -m "Branching for X.Y releases"
- Update
CHANGES.txt
to add back inTrunk (unreleased changes)
. - Update the default version in
build.xml
on trunk to X.Y+1.0-dev. - Update the
hadoop.version
number inivy/libraries.properties
on trunk to X.Y+1.0. - Commit these changes to trunk.
svn commit -m "Preparing for X.Y+1.0 development"
Updating Release Branch
These operations take place in the release branch.
- Check out the branch with:
svn co https://svn.apache.org/repos/asf/hadoop/common/branches/branch-X.Y
- Update
CHANGES.txt
to include the release version and date (this change must be committed to trunk and any intermediate branches between trunk and the branch being released). - Update
src/docs/releasenotes.html
with release notes for this release. You generate these with:edit the releasenotes.html with the list of items from $vers.html.cd src/docs jira.sh -s https://issues.apache.org/jira -u $user -p $pw \ -a getIssueList --search \ "project in (HADOOP,HDFS,MAPREDUCE) and fixVersion = '$vers' and (resolution = Fixed OR 'Target Version/s' = '$vers') ORDER BY KEY" \ | ./relnotes.py > $vers.html
- Update the version number in
build.xml
to be hadoop-X.Y.N-dev, where N is one greater than the release being made. - Update the
hadoop.version
number inivy/libraries.properties
to be the same as the release being made. - Commit these changes.
svn commit -m "Preparing for release X.Y.Z"
- If not already done, merge desired patches from trunk into the branch and commit these changes. You can find the revision numbers using
svn log CHANGES.txt
in the branch and in trunk.cd branch-X.Y svn merge -rR1:R2 ../trunk . svn commit -m "Merge -r R1:R2 from trunk to X.Y branch. Fixes: HADOOP-A, HADOOP-B."
- Tag the release candidate (R is the release candidate number, and starts from 0):
svn copy https://svn.apache.org/repos/asf/hadoop/common/branches/branch-X.Y \ https://svn.apache.org/repos/asf/hadoop/common/tags/release-X.Y.Z-rcR -m "Hadoop X.Y.Z-rcR release."
Build Requirements
To build an official release, you must:
- Use a 64-bit Linux system, so that we can build native code for both 32- and 64-bit architectures.
- Have a recent version of Eclipse installed, so that the eclipse plugin will build.
- Have Xerces C 2.8.x installed.
HADOOP-6846 has some scripts that make it easier to build and smoke test a release for 0.21.0 and later.
Building
- Build the release & run unit tests. This is captured in part in http://svn.apache.org/viewvc/hadoop/nightly/hudsonBuildHadoopRelease.sh?view=markup. The three parts of this command are intended to be run cumulatively:
## build 32-bit export JAVA_HOME=/path/to/32bit/jdk export CFLAGS=-m32 export CXXFLAGS=-m32 ant \ -Dforrest.home=/usr/local/forrest \ -Djava5.home=/usr/local/jdk1.5 \ -Dfindbugs.home=/usr/local/findbugs \ -Declipse.home=/usr/lib/eclipse \ -Dxercescroot=/usr/local/xerces-c \ -Dversion=X.Y.Z \ -Dhadoop.version=X.Y.Z \ -Dcompile.native=true \ -Dcompile.c++=true \ -Dlibhdfs=true \ -Dlibrecordio=true \ -Dtest.junit.output.format=xml \ veryclean task-controller rpm deb \ | tee build_log_dir/build32-X.Y.Z.log ## build 64-bit export JAVA_HOME=/path/to/64bit/jdk export CFLAGS=-m64 export CXXFLAGS=-m64 ant \ -Dforrest.home=/usr/local/forrest \ -Djava5.home=/usr/local/jdk1.5 \ -Dfindbugs.home=/usr/local/findbugs \ -Dversion=X.Y.Z \ -Dhadoop.version=X.Y.Z \ -Dcompile.native=true \ -Dcompile.c++=true \ -Dlibhdfs=true \ -Dlibrecordio=true \ tar rpm deb | tee build_log_dir/build64-X.Y.Z.log ## run tests (back in 32-bit mode) export JAVA_HOME=/path/to/32bit/jdk export CFLAGS=-m32 export CXXFLAGS=-m32 ant \ -Dforrest.home=/usr/local/forrest \ -Djava5.home=/usr/local/jdk1.5 \ -Dfindbugs.home=/usr/local/findbugs \ -Declipse.home=/usr/lib/eclipse \ -Dxercescroot=/usr/local/xerces-c \ -Dversion=X.Y.Z \ -Dhadoop.version=X.Y.Z \ -Dcompile.native=true \ -Dcompile.c++=true \ -Dlibhdfs=true \ -Dlibrecordio=true \ -Dtest.junit.output.format=xml \ test test-c++-libhdfs | tee build_log_dir/build32tests-X.Y.Z.log
- Check that release file looks ok - e.g. install it and run examples from tutorial.
- Generate the checksums of the release file.
gpg --print-mds hadoop-X.Y.Z.tar.gz > hadoop-X.Y.Z.tar.gz.mds
- Sign the release
gpg --armor --output hadoop-X.Y.Z.tar.gz.asc --detach-sig hadoop-X.Y.Z.tar.gz
- Copy release files to a public place.
ssh people.apache.org mkdir public_html/hadoop-X.Y.Z-candidate-0 scp -p hadoop-X.Y.Z.tar.gz* people.apache.org:public_html/hadoop-X.Y.Z-candidate-0
- Stage the release candidate to the maven repository:
For Hadoop-2.x use maven deploy
ant \ -Dforrest.home=/usr/local/forrest \ -Djava5.home=/usr/local/jdk1.5 \ -Dfindbugs.home=/usr/local/findbugs \ -Dversion=X.Y.Z \ -Dhadoop.version=X.Y.Z \ -Drepo=staging \ mvn-deploy ## Be ready to respond to the interactive request for your GPG pass-phrase, for signing the artifacts.
mvn -Psign deploy -DskipTests ## Be ready to respond to the interactive request for your GPG pass-phrase, for signing the artifacts.
- Enter Nexus, and perform the following steps:
- Click on
Log In
in the upper right corner. Log in using your apache user name and password. - In the left navigation pane, select
Staging Repositories
. - Identify the release candidate you just pushed, by your user name (in parentheses as part of the "Repository" name) and the "Created On" date. Click on the check box to the left of your Repository name to select it. (If you accidentally click on the Repository name itself, another tab will pop open. If so, just close it.)
- Click the
Close
button above the Repository names. This makes your release candidate available at the Staging level. - If you have previously staged an older release candidate with the same version number, and it is still showing in the Repository list, you must select and
Drop
the old one now. - Confirm that your new release candidate is visible at https://repository.apache.org/content/groups/staging/org/apache/hadoop/hadoop-core/X.Y.Z/ (for 1.x and 0.20.x), or https://repository.apache.org/content/groups/staging/org/apache/hadoop/hadoop-common/X.Y.Z/ (for 0.22.x, 0.23.x, and 2.x), with correct file modification dates.
- Click on
- Call a release vote on common-dev at hadoop.apache.org.
Publishing
In 7 days if the release vote passes, the release may be published.
- Tag the release:
svn move https://svn.apache.org/repos/asf/hadoop/common/tags/release-X.Y.Z-rcR \ https://svn.apache.org/repos/asf/hadoop/common/tags/release-X.Y.Z -m "Hadoop X.Y.Z release."
- Copy release files to the distribution directory and make them writable by the hadoop group.
ssh people.apache.org cp -pr public_html/hadoop-X.Y.Z-candidate-0 /www/www.apache.org/dist/hadoop/core/hadoop-X.Y.Z cd /www/www.apache.org/dist/hadoop/core chgrp -R hadoop hadoop-X.Y.Z chmod -R g+w hadoop-X.Y.Z
- The release directory usually contains just two releases, the most recent from two branches, with a link named 'stable' to the most recent recommended version.
ssh people.apache.org cd /www/www.apache.org/dist/hadoop/core rm -rf hadoop-A.B.C; rm stable ln -s hadoop-A.B.D stable
- In Nexus, promote the artifacts to 'Released' status by right-clicking the staged repository and select
Release
- Wait 24 hours for release to propagate to mirrors.
- Prepare to edit the website.
svn co https://svn.apache.org/repos/asf/hadoop/common/site ~/hadoop-site
- Update the documentation links in
.
main/author/src/documentation/content/xdocs/site.xml
- Update the release news in
.
main/author/src/documentation/content/xdocs/releases.xml
- Regenerate the site, review it, then commit it.
cd ~/hadoop-site/main ant -Dforrest.home=/usr/local/forrest -Djava5.home=/usr/local/jdk1.5 update svn commit -m "Updated site for release X.Y.Z."
- It is not usually necessary to update the site front page (http://hadoop.apache.org), but if it is needed, update
main/author/src/documentation/content/xdocs/index.xml
, then docd ~/hadoop-site/main ant -Dforrest.home=/usr/local/forrest -Djava5.home=/usr/local/jdk1.5 update svn commit -m "Updated site front page for release X.Y.Z."
- Publish the new release docs, by doing the following:
ssh people.apache.org svn co --depth immediates https://svn.apache.org/repos/asf/hadoop/common/site/main/publish/docs/ cd docs tar xzf /www/www.apache.org/dist/hadoop/core/hadoop-X.Y.Z/hadoop-X.Y.Z.tar.gz --wildcards '*/docs' mv hadoop-X.Y.Z/docs rX.Y.Z svn add rX.Y.Z svn commit -m "Publishing docs for release X.Y.Z." rm -r hadoop-X.Y.Z
- If the
docs/current
and/ordocs/stable
links should be updated to the new release, do one or both of the following:## update current rm current ln -s rX.Y.Z current svn commit -m "Updating link to current docs." ## update stable rm stable ln -s rX.Y.Z stable svn commit -m "Updating link to stable docs."
- Generate the jdiff API data for the new release by, in the branch directory, running
then commit the new XML file generated in lib/jdiff to both trunk and to the branch (and any intermediate branches between trunk and the branch being released).
ant -Dversion=X.Y.Z api-xml
svn add lib/jdiff/hadoop_X.Y.Z.xml svn commit -m "JDiff output for release X.Y.Z"
- Update the
jdiff.stable
value in the X.Y+1 branch'sbuild.xml
(which may be trunk) to be the published release (ie. X.Y.Z). - Send announcements to the user and developer lists once the site changes are visible.
- In Jira, ensure that only issues in the "Fixed" state have a "Fix Version" set to release X.Y.Z.
- In Jira, "release" the version. Visit the "Administer Project" page, then the "Manage versions" page. You need to have the "Admin" role in Hadoop Core's Jira for this step and the next.
- In Jira, close issues resolved in the release. Disable mail notifications for this bulk change.