On 3/14/07, Duncan Murdoch <murdoch@stats.uwo.ca> wrote:
> On 3/14/2007 8:21 AM, Edward Harvey wrote:
> >> -----Original Message-----
> >> From: Jared Hardy [mailto:jaredhardy@gmail.com]
> >> Sent: Tuesday, March 13, 2007 9:42 PM
> >> To: users@subversion.tigris.org
> >> Cc: Edward Harvey
> >> Subject: Re: Can I have mirrored subversion servers?
> >>
> >> The reason is: while replication/mirror solutions will
> >> certainly speed up updates (or any read-only operation),
> >> commits (write operations) should propagate to all repository
> >> servers before they are considered "complete", no matter what
> >> solution you choose.
> >
> > Well, I think that's just a matter of implementation. It is entirely
> > possible, if a system were so designed, for the LAN repository to accept
> > a commit request from the client, and quickly reserve whatever it needs
> > to reserve on all the WAN repositories before returning success.
>
> And if the LAN repository suffers a catastrophic failure (or a backhoe
> takes out their WAN connection) before the commit really finishes, the
> WAN repositories will be left waiting indefinitely, unable to process
> updates or commit, while the local user may think everything is fine.
>
> I don't know the details of what Eric hinted at for 1.5, but I hope that
> it sticks with the model of a single master repository where commits
> must succeed before a commit request returns.
See
http://svn.collab.net/repos/svn/trunk/notes/webdav-proxy
for more information.
bye,
Erik.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Wed Mar 14 14:26:59 2007