Re: any way to "re-sync" a bare repository against another bare repository?

[Michael Witten <mfwitten@gmail.com>]

On Thu, 22 Sep 2011 11:22:37 -0600, Chris Friesen wrote:

> Suppose I have a parent bare repository.  I do a "git clone --bare" to 
> create a child repository, and then clone the child to create a 
> grandchild repository.

Firstly, what exactly is the scenario you are trying to achieve? Perhaps
there is a better way to do what you are trying to do.

> If changes get pushed into the parent repository, is there any way to 
> cause the child to be updated?

The documentation answers your question (but badly, as with much of the
documentation); from `git help clone':

  --bare
      Make a bare GIT repository. That is, instead of creating
      <directory> and placing the administrative files in
      <directory>/.git, make the <directory> itself the $GIT_DIR.
      This obviously implies the -n because there is nowhere to
      check out the working tree. Also the branch heads at the
      remote are copied directly to corresponding local branch
      heads, without mapping them to refs/remotes/origin/. When
      this option is used, neither remote-tracking branches nor
      the related configuration variables are created.

In particular:

                                  Also the branch heads at the
      remote are copied directly to corresponding local branch
      heads, without mapping them to refs/remotes/origin/. When
      this option is used, neither remote-tracking branches nor
      the related configuration variables are created.

In particular:

                                                           When
      this option is used, neither remote-tracking branches nor
      the related configuration variables are created.

Thus, you have to explicitly tell git what you fetched and which
branch heads should be updated.

Consider this:

  $ git init parent
  $ git clone        parent child0
  $ git clone --bare parent child1

Now, look at the config file for the child0 repository:

  $ cat child0/.git/config 
  [core]
          repositoryformatversion = 0
          filemode = true
          bare = false
          logallrefupdates = true
  [remote "origin"]
          fetch = +refs/heads/*:refs/remotes/origin/*
          url = /path/to/parent
  [branch "master"]
          remote = origin
          merge = refs/heads/master

In particular:

          fetch = +refs/heads/*:refs/remotes/origin/*

That is the default `refspec'; when `git fetch' is not explicitly
told on the command line what to fetch and which branch head[s] to
update, then this refspec is used as a default.

Now, look at the config file for the child1 repository:

  $ cat child1/config
  [core]
          repositoryformatversion = 0
          filemode = true
          bare = true
  [remote "origin"]
          url = /path/to/parent

In particular, note that a bare repository doesn't include any
such default information for `git fetch'. However, you could be
explicit about it; from within the chidl1 repo:

  $ git fetch origin '+refs/heads/*:refs/remotes/origin/*'

> Just a "git fetch <parent>" doesn't seem to help.  If I set up parent as 
> a remote branch I can fetch it,

Firstly, it doesn't make any sense to say "set up parent as a remote
branch"; what you mean is "set up `<parent>' as a remote with a default
refspec that creates any associated remote-tracking branch heads".

Secondly, by setting up `<parent>' as a remote, you are creating the
missing refspec in your config file:

  [remote "<parent>"]
          url = /path/to/parent
          fetch = +refs/heads/*:refs/remotes/<parent>/*

which is why you get this:

> but then it shows all the branches as "<parent>/<branch>" rather
> than updating the child.

You need a different refspec, namely:

  +refs/heads/*:refs/heads/*

So, either be explicit:

  git fetch '<parent>' '+refs/heads/*:refs/heads/*'

or update your config:

  git config 'remote.<parent>.fetch' '+refs/heads/*:refs/heads/*'

Of course, there is an easier way that does all of this [and more]
for you:

> I just tried a "git clone --mirror" to create the child and it seems to 
> allow me to pick up changes in the parent via "git fetch".  Is that the 
> proper way to handle this?

The documentation answers your question (but badly, as with much of the
documentation); from `git help clone':

  --mirror
      Set up a mirror of the source repository. This implies
      --bare. Compared to --bare, --mirror not only maps local
      branches of the source to local branches of the target, it
      maps all refs (including remote-tracking branches, notes
      etc.) and sets up a refspec configuration such that all
      these refs are overwritten by a git remote update in the
      target repository.

In particular:

                sets up a refspec configuration such that all
      these refs are overwritten by a git remote update in the
      target repository.

Consider this:

  $ git clone --mirror parent child2
  $ cat child2/config
  [core]
          repositoryformatversion = 0
          filemode = true
          bare = true
  [remote "origin"]
          fetch = +refs/*:refs/*
          mirror = true
          url = /path/to/parent

In particular:

          fetch = +refs/*:refs/*

That's a very liberal refspec! It basically says that fetch
should mirror everything by default.

Sincerely,
Michael Witten