Re: Question re. git remote repository

From: Jeff King <peff@peff.net>
Date: Jan 17, 2013, 23:49 UTC
Message-ID: <20130117234912.GA14936@sigill.intra.peff.net>
In-Reply-To: <alpine.DEB.2.02.1301161459060.21503@nftneq.ynat.uz>

On Wed, Jan 16, 2013 at 03:00:41PM -0800, David Lang wrote:

Show 16 quoted lines

> >>This one [1] for instance.  I also recall seing people having other
> >>"mystical" problems with setups like this so I somehow developed an idea
> >>than having a repository on a networked drive is asking for troubles.
> >>Of course, if there are happy users of such setups, I would be glad to
> >>hear as my precautions might well be unfounded for the recent versions
> >>of Git.
> >>
> >>1. http://code.google.com/p/msysgit/issues/detail?id=130
> >
> >A group I was with used a master repository on a windows share for quite some time without a database corruption being seen.   --
> 
> I think the risk is that if you have multiple people doing actions on
> the shared filesystem you can run into trouble.
> 
> As long as only one copy of git is ever running against the
> repository, I don't see any reason for there to be a problem.

That should not be an issue. Git on a server has to deal with multiple independent receive-pack's running to accept several simultaneous pushes. They coordinate through the use of file locks. Having multiple machines pushing over a shared filesystem should work the same, as long as the filesystem support atomic creation of files with O_EXCL.

There may be other subtle issues lurking (e.g., the Windows issue that Konstantin mentioned), but as far as I know, it _should_ work in general.

-Peff