Add script to semi-automate bulk releases

The release team has a few cases now where we need to propose release
patches for a large set of deliverables. Due to differences in semver
choice based on actually merged commits, along with other decisions that
need to be made while doing these, this process can't be completely
automated. But this adds a script that will automate the majority of the
process to simplify it as much as possible.

Change-Id: I6ec9fa77baab58df93bdadc0ac3c3fa5d3e18804
Signed-off-by: Sean McGinnis <sean.mcginnis@gmail.com>
This commit is contained in:
Sean McGinnis 2020-06-15 14:34:46 -05:00
parent 5617dcc77f
commit 355cd7debd
No known key found for this signature in database
GPG Key ID: CE7EE4BFAF8D70C8
2 changed files with 192 additions and 6 deletions

View File

@ -679,13 +679,14 @@ the PTL and all release liaisons.
This is designed to be used by the release team at key points in the cycle to
ease bulk releases.
.. note::
.. note::
This tool will commit ultimately commit all modified deliverables and
modifies git state. Therefore it is essential that before running it
the working tree contains only the logical changes appropriate for the
stage of the release *and* all changes are saved elsewhere, in case the
script encounters a problem.
This tool will commit ultimately commit all modified deliverables and
modifies git state. Therefore it is essential that befoer running it
the working tree contains only the logical changes appropriate for the
stage of the release *and* all changes are saved elsewhere, in case the
script encounters a problem.
tools/make_missing_releases.sh
------------------------------
@ -695,6 +696,41 @@ to create releases at appropriate times in the release cycle, e.g milestones.
Once ``tools/make_missing_releases.sh`` completes the release manager can use
``tools/bulk_review.sh`` to submit the release requests.
toos/process_auto_releases.sh
-----------------------------
Automates parts of the process to propose releases for a large set of
deliverables.
There are multiple points during the release cycle where the release team
needs to initiate releases for library releases, tagging RCs, or other cases
where we need to inspect each deliverable in a set to generate release
requests.
This tool asks for input on a few common settings to use for the releases. A
template commit message is entered on the command line, using the placeholder
of PROJECT that will be replaced by the actual deliverable name. It then
iterates through a set of deliverables and shows any commits to the relevant
repos that have not been included in a release yet. You are then able to
decide whether it needs to be released and select the release type
(major, minor, bugfix, rc, etc.) based on the included commits.
This can be used in conjunction with the ``list-deliverables`` command to get
the specific deliverables to process. An example use would be::
./tools/process_auto_releases.sh ussuri $(list-deliverables --unreleased --series ussuri)
As an alternative, it may be useful to be able to edit the list of deliverables
before running the command. That can be done by something similar to::
tox -e venv -- list-deliverables --unreleased --series ussuri > deliverables.log
vi deliverables.log # edit contents as needed
./tools/process_auto_releases.sh ussuri $(cat deliverables.log)
Unlike make_missing_releases.sh, this script will create fresh temporary clones
of each repo to avoid stale information, and it will submit each new release
request as it goes.
tools/releases_note_links.sh
------------------------------

150
tools/process_auto_releases.sh Executable file
View File

@ -0,0 +1,150 @@
#!/bin/bash
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
#
# Semi-automates the process of creating release team initiated release
# requests.
if [[ $# -lt 2 ]]; then
echo "Usage: $(basename $0) <series> <repo> [<repo>...]"
echo "repo should be e.g. glance"
echo
echo "Example: $(basename $0) ussuri \$(list-deliverables --unreleased --series ussuri)"
echo "Example: $(basename $0) train \$(list-deliverables \\"
echo " --series train --cycle-based-no-trialing | grep -v tempest)"
echo "Example: $(basename $0) train \$(list-deliverables --series stein --cyle-based)"
exit 1
fi
series="$1"
shift
repos="$@"
TOOLSDIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
BASEDIR=$(dirname $TOOLSDIR)
source $TOOLSDIR/functions
# Make sure no pager is configured so the output is not blocked
export PAGER=
if [[ -z "$VIRTUAL_ENV" ]]; then
if [[ ! -d .tox/venv ]]; then
tox -e venv --notest
fi
source ./.tox/venv/bin/activate
fi
# Set up a clean workspace so we make sure we always have the latest and clean
# files for processing. setup_temp_space defined $MYTMPDIR.
setup_temp_space 'make-auto-releases'
clone_repo "openstack/releases"
cd openstack/releases
git review -s > /dev/null
# Set the branch name for this series
branch=$(series_to_branch "$series")
# Prompt for global parameters to use for all releases
read -p "> Enter review topic to use: " topic
echo
echo "> Enter commit message template"
echo " The placeholder PROJECT will be replaced with the current project."
echo " (Press Ctrl-D when done.)"
echo
echo "------------------------------------------------------------------------"
commit=$(</dev/stdin)
echo
newbranch=""
read -p "> Create stable branch? [y/N]: " YN
if [ "${YN,,}" == "y" ]; then
newbranch="--stable-branch"
fi
echo
echo "======================================"
echo "Confirm details:"
echo "======================================"
echo "Topic: $topic"
echo "Series: $series"
echo "Branch: $branch"
echo "New Branch: $newbranch"
echo "Commit Message Template:"
echo
echo "------------------------------------------------------------------------"
echo "${commit//PROJECT/example-project}"
echo "------------------------------------------------------------------------"
echo
echo "======================================"
read -p "Continue with processing? [y/N]: " YN
if [ "${YN,,}" != "y" ]; then
exit 1
fi
function process_repo {
repo=$1
title "Unreleased changes in $repo ($series)"
cd "$MYTMPDIR"
clone_repo "openstack/$repo" $branch
if [[ $? -ne 0 ]]; then
return 1
fi
cd "openstack/$repo"
prev_tag=$(get_last_tag)
if [ -z "$prev_tag" ]; then
echo "$repo has not yet been released"
else
echo
end_sha=$(git log -n 1 --pretty=tformat:%h)
echo "Changes between $prev_tag and $end_sha"
echo
git log --no-color --no-merges --format='%h %ci %s' \
--graph ${prev_tag}..${end_sha}
echo
fi
read -p "Create new release? [y/N]: " YN
case $YN in
[Yy]* ) RELEASE="y";;
* )
echo "Skipping $repo release..."
return
;;
esac
echo
echo "Releasing $repo"
echo "===================="
cd ../releases
echo "Select appropriate release type:"
select type in bugfix feature major milestone rc
do
new-release $series $repo $type $newbranch
break
done
git add .
message=${commit//PROJECT/$repo}
git commit -s -m "$message"
git log -1
git review -t $topic
git reset --hard HEAD~1 > /dev/null
cd ../..
echo
}
# Process each repo passed in to see if a release should be proposed
for repo in $repos; do
cd $MYTMPDIR
echo
process_repo "${repo/openstack\//}"
done