|
From: | Ted Timmons |
Subject: | Re: [Duplicity-talk] large upload "ETA stalled"? |
Date: | Mon, 16 Jan 2017 16:54:14 +0000 |
i wonder if it'd make sense to somehow mark/tag backends that support progress and throw an error when --progress is enabled on a backend that currently does not support it?
local file progress should probably properly implement via
http://bazaar.launchpad.net/~duplicity-team/duplicity/0.8-series/view/head:/duplicity/path.py#L622
to avoid "stalling" on local targets. users might be confused when their otherwise outperforming system claims to be "stalling".
..ede/duply.net
On 16.01.2017 03:08, Cláudio Gil wrote:
> The previous branch now includes the "autoprogress" decorator and uses it
> in the b2 backend. I've also added it to the local backend.
>
> I've done a few tests with the local backend and it works as expected.
> Since it only provides progress at the end of the upload/download then it
> still shows stalled when volumes take longer than 5s (by default) to
> transfer. Using "--progress-rate=<sec>" can avoid that. For example, with
> the default volume size of 200MB and --progress-rate=60 you will see
> "Stalled", one in a while, when the bandwidth is bellow 1.7MB/s (bandwidth
> = volsize/(rate*2)).
>
> Cheers,
> Cláudio
>
> 2017-01-15 12:46 GMT+00:00 <address@hidden>:
>
>> hey Claudio,
>>
>> if it's really that easy you might consider writing a decorator that can
>> be added easily to the remaining backends, that do not need special
>> handling.
>>
>> ..ede/duply.net
>>
>> On 15.01.2017 13:41, Cláudio Gil wrote:
>>> The boto API supports a progress callback which means you can get
>> progress updates during the upload. Dropbox API also supports
>> chunks/appends but I don't think other APIs do. I haven't really looked
>> into it.
>>>
>>> In any case, you are right, the code only introduces progress report for
>> the upload part. It can be easily added to the get (and other backend)
>> because it is a very rough progress report which don't requires any special
>> support.
>>>
>>> Cheers,
>>> Cláudio
>>>
>>> edgar.soldin--- via Duplicity-talk <address@hidden <mailto:
>> address@hidden>> escreveu em dom, 15/01/2017 às 11:30 :
>>>
>>> On 15.01.2017 12:15, Cláudio Gil via Duplicity-talk wrote:
>>>
>>> > Hi,
>>>
>>> >
>>>
>>> > https://code.launchpad.net/~m4ktub/duplicity/0.7-fixes/+
>> merge/314778 <https://code.launchpad.net/%7Em4ktub/duplicity/0.7-fixes/+
>> merge/314778> <https://code.launchpad.net/%7Em4ktub/duplicity/0.7-fixes/+
>> merge/314778>
>>>
>>> > If you look at the diff you can see that I've split the import
>> into two. The rest remains the same.
>>>
>>> >
>>>
>>> > This kind of progress reporting does not require any support from
>> the backend's API so it would work for all other backends, I suppose. I
>> don't use progress but if others feel important I can try to add this kind
>> of rough progress into the other backends. Maybe this would be 0.8-series
>> material...
>>>
>>> >
>>>
>>>
>>>
>>> the progress reporting in _boto_{multi,single}.py looks a bit more
>> complex than that.
>>>
>>>
>>>
>>> is it possible that this fix only handles the progress for uploads
>> but ignores downloads _get() ?
>>>
>>>
>>>
>>> ..ede/duply.net <http://duply.net>
>>>
>>>
>>>
>>> _______________________________________________
>>>
>>> Duplicity-talk mailing list
>>>
>>> address@hidden <mailto:address@hidden>
>>>
>>> https://lists.nongnu.org/mailman/listinfo/duplicity-talk
>>>
>>
>
_______________________________________________
Duplicity-talk mailing list
address@hidden
https://lists.nongnu.org/mailman/listinfo/duplicity-talk
[Prev in Thread] | Current Thread | [Next in Thread] |