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_OK
Now 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_OK
Git 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.
References
No webmentions were found.