N-way Git synchronization with extra cheese
Index
- Background
- Converting Subversion to Git
- Generate and version .gitignore files
- Git via proxy
- Setting up pull everywhere
- References
Background
I've got a desktop and server behind a router with a dynamic IP address at home, a desktop at work, and a laptop that floats around. I'd very much like to have the same settings on all of them, and to be able to synchronize them as easily as possible. I've been using Subversion for this, but recent trouble with symlinks and a long-term concern that storing the revision history centrally (even with backups now and then) is a Bad Move in the long term. So when I had to start using Git at work, and after realizing that it could solve both problems (at least in theory), I tried figuring out how to do this. After lots of tries followed by rm -rf settings/, I think I've got a working setup. Of course, I don't guarantee that any of this will work for you.
Converting Subversion to Git
Install the necessary software:
sudo apt-get install git-svn
Copy the following code into a file named svn2git.sh, and run it as documented below.
svn2git.sh
#!/bin/sh
#
# NAME
# svn2git.sh - Convert a Subversion repository to Git
#
# SYNOPSIS
# svn2git.sh [options] <Subversion URL>
#
# OPTIONS
# --authors=path Authors file
# -v,--verbose Verbose output
#
# EXAMPLE
# /path/to/svn2git.sh https://example.org/foo
#
# Create authors file for repository
#
# /path/to/svn2git.sh -v --authors=authors.txt https://example.org/foo
#
# Get Subversion repository to ./foo.git
#
# DESCRIPTION
# Two-part script to migrate from Subversion to Git. First it tries to get
# a list of the Subversion authors, so it can be formatted to fit the Git
# commit structure. When running with the authors file, it will fetch the
# entire Subversion revision history.
#
# BUGS
# Email bugs to victor dot engmark at gmail dot com. Please include the
# output of running this script in verbose mode (-v).
#
# COPYRIGHT AND LICENSE
# Copyright (C) 2009 Victor Engmark
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
################################################################################
# Output error message with optional error code
error()
{
if [ -z "$2" ]
then
error_code=$EX_UNKNOWN
else
error_code=$2
fi
echo "$1" >&2
exit $error_code
}
usage()
{
error "Usage: ${cmdname} [-v|--verbose] [--authors=path] <Subversion URL>" $EX_USAGE
}
verbose_echo()
{
if [ $verbose ]
then
echo "$*"
fi
}
# Use for mandatory directory checks
# $1 is the directory path
# $2 is the (optional) error message
directory_exists()
{
if [ ! -d $1 ]
then
error "No such directory '${1}'
$2" $EX_NO_SUCH_DIR
fi
}
# Make sure an executable is available
# $1 is the path to the executable
# $2 is the (optional) error message
executable_exists()
{
if [ ! -x $1 ]
then
error "No such executable '${1}'
$2" $EX_NO_SUCH_EXEC
fi
}
PATH="/usr/bin:/bin"
cmdname=`basename $0`
directory=$PWD
# Exit codes from /usr/include/sysexits.h, as recommended by
# http://www.faqs.org/docs/abs/HTML/exitcodes.html
EX_OK=0 # successful termination
EX_USAGE=64 # command line usage error
EX_DATAERR=65 # data format error
EX_NOINPUT=66 # cannot open input
EX_NOUSER=67 # addressee unknown
EX_NOHOST=68 # host name unknown
EX_UNAVAILABLE=69 # service unavailable
EX_SOFTWARE=70 # internal software error
EX_OSERR=71 # system error (e.g., can't fork)
EX_OSFILE=72 # critical OS file missing
EX_CANTCREAT=73 # can't create (user) output file
EX_IOERR=74 # input/output error
EX_TEMPFAIL=75 # temp failure; user is invited to retry
EX_PROTOCOL=76 # remote error in protocol
EX_NOPERM=77 # permission denied
EX_CONFIG=78 # configuration error
# Custom errors
EX_UNKNOWN=1
EX_NO_SUCH_DIR=91
EX_NO_SUCH_EXEC=92
# Process parameters
until [ $# -eq 0 ]
do
case $1 in
-v|--verbose)
verbose=1
shift
;;
--authors=*)
authors_file=${directory}/$(echo "$1" | cut -c11-)
shift
;;
*)
if [ -z $svn_url ]
then
svn_url=$1
shift
else
# Unknown parameter
usage
fi
;;
esac
done
if [ -z $svn_url ]
then
# No Subversion URL provided
usage
fi
repository_name=`basename $svn_url`
verbose_echo "Running $cmdname at `date`."
# Preliminary checks
directory_exists "$source_base"
executable_exists "/usr/bin/git"
executable_exists "/usr/bin/git-svn"
executable_exists "/usr/bin/svn"
verbose_echo "Source repository: '${svn_url}'"
if [ -z $authors_file ]
then
# Get authors file
authors_file="${directory}/${repository_name}-authors.txt"
if [ -e $authors_file ]
then
error "Authors file '${authors_file}' already exists"
fi
verbose_echo "Authors file: ${authors_file}"
svn log --quiet "${svn_url}" | grep '^r.*' | cut -d ' ' -f 3- | cut -d '|' -f 1 | sort | uniq > "${authors_file}"
author="$(head -1 $authors_file)"
echo "Please modify ${authors_file} to a format like"
echo "${author}= Full Name <${author}@example.org>"
echo "and rerun $cmdname with --authors=${authors_file}"
else
if [ ! -e $authors_file ]
then
error "Authors file '${authors_file}' doesn't exist"
fi
git_target="${directory}/${repository_name}.git"
if [ -e $git_target ]
then
error "Target repository '${git_target}' already exists"
fi
verbose_echo "Target repository: '${git_target}'"
# Clone
git-svn clone --no-metadata --authors-file="${authors_file}" --revision 1:1 "$svn_url" "$git_target" || error "Clone failed"
# Fetch
cd "$git_target"
batch_start=2
revisions=$(svn info "$svn_url" | grep '^Revision:' | awk '{print $2}')
while [ $batch_start -le $revisions ]
do
batch_end=$(expr $batch_start + 990)
if [ $batch_end -gt $revisions ]
then
batch_end=$revisions
fi
verbose_echo "Fetching revisions $batch_start through $batch_end"
git-svn fetch --authors-file="${authors_file}" --revision $batch_start:$batch_end || error "Fetch failed"
batch_start=$(expr $batch_end + 1)
done
git rebase git-svn
verbose_echo "Applying svn:ignore properties"
git-svn show-ignore >> .git/info/exclude
verbose_echo "Removing references to Subversion"
git config --remove-section svn-remote.svn
rm --recursive --force .git/svn/
fi
verbose_echo "Cleaning up."
cd "$directory"
verbose_echo "${cmdname} completed at `date`."
exit $EX_OKNow make sure you do a directory diff between the old Subversion and the new Git repositories to see if it succeeded.
Now you can get this on other machines using
git clone --origin example ssh://example.org/~/settings
Generate and version .gitignore files
This is an optional step in case you would like to version the old svn:ignore properties as .gitignore files:
exclude2gitignore.sh
#!/bin/sh
#
# NAME
# exclude2gitignore.sh - Convert $GIT_DIR/info/exclude to corresponding
# .gitignore files
#
# SYNOPSIS
# exclude2gitignore.sh [options] /path/to/repository
#
# OPTIONS
# -v,--verbose Verbose output
#
# EXAMPLE
# /path/to/exclude2gitignore.sh ~/foo
#
# Create .gitignore files for the Git repository in ~/foo
#
# DESCRIPTION
# Based on the format generated by `git-svn show-ignore`, where non-comment
# lines indicate ignored files. Will try to put the .gitignore as close as
# possible to the ignored file(s).
#
# BUGS
# Email bugs to victor dot engmark at gmail dot com. Please include the
# output of running this script in verbose mode (-v).
#
# COPYRIGHT AND LICENSE
# Copyright (C) 2009 Victor Engmark
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
################################################################################
# Output error message with optional error code
error()
{
if [ -z "$2" ]
then
error_code=$EX_UNKNOWN
else
error_code=$2
fi
echo "$1" >&2
exit $error_code
}
usage()
{
error "Usage: ${cmdname} [-v|--verbose] /path/to/repository" $EX_USAGE
}
verbose_echo()
{
if [ $verbose ]
then
echo "$*"
fi
}
# Use for mandatory directory checks
# $1 is the directory path
# $2 is the (optional) error message
directory_exists()
{
if [ ! -d $1 ]
then
error "No such directory '${1}'
$2" $EX_NO_SUCH_DIR
fi
}
# Make sure an executable is available
# $1 is the path to the executable
# $2 is the (optional) error message
executable_exists()
{
if [ ! -x $1 ]
then
error "No such executable '${1}'
$2" $EX_NO_SUCH_EXEC
fi
}
PATH="/usr/bin:/bin"
cmdname=`basename $0`
directory=$PWD
# Exit codes from /usr/include/sysexits.h, as recommended by
# http://www.faqs.org/docs/abs/HTML/exitcodes.html
EX_OK=0 # successful termination
EX_USAGE=64 # command line usage error
EX_DATAERR=65 # data format error
EX_NOINPUT=66 # cannot open input
EX_NOUSER=67 # addressee unknown
EX_NOHOST=68 # host name unknown
EX_UNAVAILABLE=69 # service unavailable
EX_SOFTWARE=70 # internal software error
EX_OSERR=71 # system error (e.g., can't fork)
EX_OSFILE=72 # critical OS file missing
EX_CANTCREAT=73 # can't create (user) output file
EX_IOERR=74 # input/output error
EX_TEMPFAIL=75 # temp failure; user is invited to retry
EX_PROTOCOL=76 # remote error in protocol
EX_NOPERM=77 # permission denied
EX_CONFIG=78 # configuration error
# Custom errors
EX_UNKNOWN=1
EX_NO_SUCH_DIR=91
EX_NO_SUCH_EXEC=92
# Process parameters
until [ $# -eq 0 ]
do
case $1 in
-v|--verbose)
verbose=1
shift
;;
*)
if [ -z $repository ]
then
repository="${1%\/}"
shift
else
# Unknown parameter
usage
fi
;;
esac
done
verbose_echo "Running $cmdname at `date`."
directory_exists "$repository"
grep '^/' "${repository}/.git/info/exclude" | while read line
do
ignore_path="${repository}${line}"
verbose_echo "Starting with $ignore_path"
ignore_name="$ignore_path"
# Strip globs in path
ignore_path=`dirname "$ignore_path"`
while [ ! -e "$ignore_path" ]
do
ignore_path=`dirname "$ignore_path"`
done
# Remove path from file name (need +2 to include the end slash and to
# compensate for 1-based indexing
name_length=$(expr length "$ignore_name")
path_length=$(expr length "$ignore_path" + 2)
ignore_name=$(expr substr "$ignore_name" $path_length $name_length)
# Complete .gitignore path
ignore_path="${ignore_path}/.gitignore"
verbose_echo "$ignore_name >> $ignore_path"
echo "$ignore_name" >> "$ignore_path"
done
verbose_echo "Cleaning up."
cd "$directory"
verbose_echo "${cmdname} completed at `date`."
exit $EX_OKGit via proxy
One of the machines involved is behind a gateway machine at work, so I had to add the following to ~/.ssh/config:
Host work
ProxyCommand ssh -q gateway.example.org nc %h %p $*
HostName work-pc.example.org
With this, it's possible to refer to just "work", and SSH commands (even via Git) will take care of connecting via the proxy.
Setting up pull everywhere
The main idea here is to set up Git "remotes" pointing to all the other machines.
To be able to get the updates from the repository in ~/settings on my.example.org, simply run the following on all machines (except, of course, the home machine):
git remote add home ssh://home-pc.example.net/~/settings
To be able to get the updates from the "work" host specified with a proxy above, just use "work" for the host name:
git remote add work ssh://work/~/settings
To be able to pull from a machine which changes IP address, you could set up a DynDNS account and use one of their recommended update scripts to be able to refer to your machine using a single DNS name.
After cloning one of the copies on all of your hosts, you should be able to do the following to get all the changes from the repositories:
git remote update && git pull
If this doesn't work, you might have more luck fetching each repository individually, and then rebasing to it:
git fetch home && git rebase home/master
To keep a backup on a separate machine, just do a
git clone --origin example ssh://example.org/~/settings
there and set up pushing defaults on the other machines using
git config push.default matching
git remote add backup ssh://backup.example.org/~/settings
Then you can just git push backup master to backup the local master branch.