Re: [Duplicity-talk] Anyone using duplicity with box.net (webdavs)?

[edgar . soldin]

On 03.03.2012 05:02, Michael Terry wrote:
> On 29 February 2012 11:14,  <address@hidden> wrote:
>> On 29.02.2012 17:05, Ken Bass wrote:
>>> Not sure if it is related to the allprop thing, but when I tried using 
>>> davfs to mount box.net (webdavs) as a filesystem, the problem was that
>>> even though a file would be uploaded, it would either disappear from the 
>>> filesystem or show a size of zero. I think there was some sort of lag
>>> between the file being committed to the server and showing up on the local 
>>> davfs mounted filesystem. If I recall that caused duplicity to fail because 
>>> it would query the file it just uploaded and either the size would be 
>>> returned incorrectly or the file would appear missing.
>>
>> yeah we have added such a check lately, maybe we should have a switch to 
>> disable it for corner cases like these.
>> @mt: what do you think? btw. sorry, haven't gotten around to implement it 
>> further so far ;( but i am still aware :)
> 
> Well, I'd prefer we were able to fix it internally rather than
> introduce a user-driven workaround for a problem the user shouldn't be
> expected to understand.
> 
> Maybe just turn off the check for webdav if it's going to be a
> persistent problem.
> 
> Or webdav could add a sleep delay to let the server catch up?  Or
> maybe there's a protocol way of waiting for confirmation that the file
> is there?
> 


in this case the problem is either the webdav service provider or how davfs is 
handling the folder listing.

generally it'd make sense to have the plausibility size check retry if the file 
was missing, just for cases like these. i guess the cloud nature of the service 
adds a delay to the availability of uploaded data.

webdav has no way of "protocol way of waiting for confirmation that the file is 
there" as far as i checked the specs.
http://www.webdav.org/specs/#dav

..ede