Ok, but that's not what i we at work. The local svn will be a "working copy"
of the external svn. In fact, no user have to use the external svn if
there's a local svn. The goal of this is the limit the access to the WAN. If
the external svn is sync twice per day, we can limit these access and
decrease the commit time for internal user (we have no external user). A LAN
commit is faster than a WAN commit (the bandwith is lower for wan) And we
can have a better security, we have no control on what's do on the WAN with
our data and many commit from LAN to WAN are "point of access to private
data" (SSL or not, ssl is not a good security, you can spoof him easily). We
can limit this risk by limitting the data who are sent.
-----Message d'origine-----
De : Ryan Schmidt [mailto:subversion-2010b_at_ryandesign.com]
Envoyé : lundi 2 août 2010 12:04
À : Istace Emmanuel
Cc : 'Tony Sweeney'; users_at_subversion.apache.org
Objet : Re: SVN "Relay"
On Aug 2, 2010, at 04:54, Istace Emmanuel wrote:
> Ryan Schmidt wrote:
>> Yes, "svnsync" is the software you are looking for. You can read all
>> about it in the book:
>>
>> http://svnbook.red-bean.com/nightly/en/svn.reposadmin.maint.ht
>> ml#svn.reposadmin.maint.replication
>>
>> You will probably want to sync constantly, not just twice a day.
>
> Tony Sweeney wrote:
>> Actually, I think he's looking at something more along the lines of
>> Perforce's P4Proxy server, but for Subversion.
>>
>> http://www.perforce.com/perforce/doc.current/manuals/p4sag/09_p4p.htm
>> l
>>
>> Svnsync doesn't help in the event that someone not on his LAN
>> independently commits to the Subversion server on the WAN, as you'd
>> need to be able to sync bidirectionally, and I'm not even sure that's
possible.
>
> To Ryan : Thanks a lot for this, you will make my router happy :D For
> the replication i will see what's the traffic generated by a constant
> réplication. This WAN svn service is a temp service, in the next week
> we will host everything locally and just keep an external svn backup
server.
>
> To Tony : It's a possibility but i don't have to do a bidirectionnal
sync.
> But i will have a look at your solution and choose between the two after.
There shouldn't be much difference in bandwidth whether you svnsync twice a
day or immediately after every revision is committed; svnsync replays the
commits in order so they're going to be the same size either way.
No users will be writing to the LAN mirror; all users will be writing to the
master repository on the WAN, even those users who checked out from the LAN
mirror. This is accomplished by setting up a write-through proxy.
Received on 2010-08-02 12:17:13 CEST