[Dirvish] Dirvish backup controlled by client?

[Dave Howorth]

Noel Kelly wrote:
> Dave Howorth wrote:
>> Frank Lüken wrote:
>>> Hi!
>>>
>>> Can i config dirvish that way, that i can initialize the backup from
>>> the client-pc?
>>>
>>> Still i use a Dirvish backup server which connect to some client pc's
>>> and start the backup process. But in my network are notebooks too and
>>> these notebooks have a dynamic ip and name. So i want to configure
>>> dirvish at each notebook and than the users should be able to run the
>>> backup by themselves. Can i do this and if yes, how must i configure
>>> dirvish?
>> You can do it by writing a shell script that you invoke on the server
>> from the client via ssh. The shell script should run dirvish on the
>> appropriate vault. This means your client machines have login access to
>> your server to invoke a root process. That can cause security problems
>> unless done very carefully. I don't believe dirvish offers the facility
>> directly, for that reason.
> 
> I guess the problem Frank has is that the server won't know the IP of 
> the client it has to talk to as he said the hostnames and IPs are all 
> dynamic.  There must be a way of passing this though but not sure how to 
> get that into dirvish.

Hmm, hadn't spotted that :(

I have trouble understanding why each laptop, or user, or whatever
entity actually 'owns' the dataset doesn't/can't have a persistent name?
If there's a persistent dataset that needs to be backed up it'd be most
odd if it couldn't be named (by a file in a well-known location, for
example)

Given that, then give the laptops a [auxiliary] DNS mapping that locates
the dataset and use that in scripts, dirvish config etc.

I suspect the real problem is actually simpler than described :) I guess
Frank will describe it in more detail if he hasn't seen a solution yet!

Cheers, Dave