[git-users] git backup

2013-10-15 Thread shlo . afgin


Hi,
All the git repositories are on one git server in specific directory.
I would like to know what I need to backup in order to be able to make a 
success restore, in case my  disk  on the git server will crushed.

I would like to understand, after the user run on his local workstation 
'git init / add / commit / push'  all the information was copied to the git 
server and the local directory in the local workstation can be removed? 

I can see  somehow the name of the files and their content in the git 
repository that saved on the git server?
Thanks.

-- 
You received this message because you are subscribed to the Google Groups Git 
for human beings group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.


Re: [git-users] git backup

2013-10-15 Thread Konstantin Khomoutov
On Tue, 15 Oct 2013 06:53:17 -0700 (PDT)
shlo.af...@gmail.com wrote:

 All the git repositories are on one git server in specific directory.
 I would like to know what I need to backup in order to be able to
 make a success restore, in case my  disk  on the git server will
 crushed.
 
 I would like to understand, after the user run on his local
 workstation 'git init / add / commit / push'  all the information was
 copied to the git server and the local directory in the local
 workstation can be removed? 

Depends on how you define all the information: contrary to
centralized systems, in Git, pushing is a voluntary act -- just like
sharing, -- where the one who shares decides what they want to share.
In the case of Git, a developer chooses which bits of their local
commit graphs to send, and which objects to update with them.

Specifically:
* If you have recorded 10 commits on top of your local branch X which
  tracks a remote branch, you're not required to update the branch being
  tracked with all those 10 commits -- you might send, say, just the
  first three of them.
* Tags are not pushed unless `git push` is explicitly told to do that,
  and the tags to send are specified.  Notes are not sent as well.
* New local branches are not pushed unless `git push` is told to do so.

On the other hand, what's sent, is sent completely and with all the
dependent objects, so if someone pushes a commit into the repository,
you can be sure everything this commit references in the repository it
has been created in, is also sent.

Note that what I have just described -- the freedom of the owner of
a local repository in what they share -- is just a property of a DVCS
you have to live with.  But what ends up in your centralized (shared)
repository is self-consistent, and you can safely back it up.

What's pushed when a developer simply runs `git push` or
`git push remote` in their repository is another story completely as
it depends on a number of details.  If you're interested about this, ask
away.

 I can see  somehow the name of the files and their content in the git 
 repository that saved on the git server?

Yes: the regular Git commands for inspecting the commit graphs work in
bare repositories just as well as they do in normal ones.

-- 
You received this message because you are subscribed to the Google Groups Git 
for human beings group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.