Re: [BackupPC-users] pre-load backup

[Les M. <les...@gm...>]

On 11/12/10 3:27 AM, McDonagh, Ed wrote:
>> -----Original Message-----
>> From: Luis Paulo [mailto:lui...@gm...]
>> Sent: 12 November 2010 02:49
>> To: General list for user discussion,questions and support
>> Subject: Re: [BackupPC-users] pre-load backup
>>
>> On Fri, Nov 12, 2010 at 01:39, higuita<hi...@gm...>  wrote:
>>> Hi
>>>
>>>         i have a machine that have about 100GB of data to backup, via
>>> ssh+rsync, but its network connection is about 1Mb, so it will take ages
>>> to do even the first backup.
>>>
>>>         i already have on the backuppc server one old backup of
>>> that machine, done via plain rsync (not backuppc).
>>>
>>
>> I believe you may:
>> 1- configure backup to make a backup of the 100MB on the server. Do it once.
>> 2- after that when you run the client machine's backup, files will be
>> already on the pool (or cpool), and backup will be faster, smaller
>> transfer.
>>
>> Yet to prove, though.
>>
>> Regards
>> Luis
>
>
> I think I am right in saying that this will only work if backuppc thinks that both the local server and the remote one are the same server. This is because backuppc/rysnc will only transfer files that are different from the version it already has _for that machine_. Once it has transferred the files, it later does a check against the files it already has in pool or cpool, and if a duplicate exists it creates a hard link and deletes the new copy.
>
> So just by having the files already in pool/cpool, you won't avoid transferring anything.
>
> What you would need to do is set it up as if for the remote host, then use the alias configuration option to point it at the local one. Then once you have done a full backup, change the alias back to the remote host.
>
> I think this is right, but it is only conjecture from reading previous posts on here, so perhaps someone with better knowledge can confirm or correct me!

Yes, the catch is that you have to make the files appear in the same location in 
the temporary host for the run and use the ClientAlias setting to make the 
backup access this host instead of the real target.  Once the initial copy is in 
place you can remove the ClientAlias and pick up with the differences from the 
real host.

-- 
   Les Mikesell
    les...@gm...