Lars Wirzenius <l...@liw.fi> writes:
> On Fri, Nov 27, 2015 at 11:14:34AM +0100, Ansgar Burchardt wrote:
>> I'm using Obnam to create a full backup of my ${HOME} to a local disk
>> attached via USB3. During this obnam seems to create checkpoints quite
>> often, so that just that seems to take half of the time:
> ...
>> The "re-opening repository" step takes around 30s after which 30s
>> writing data follows, then the next checkpoint happens.
>
> There's no checkpoint value that is good for everyone, I think. The
> balance between the time between checkpoints and the time it takes to
> commit a checkpoint generation and start a new one is difficult to get
> right automatically, as it depends on various environmental factors,
> such as speed and reliability of the connection to the backup
> repository. The default checkpoint size is chosen for semi-unreliable,
> relatively slow connections.
>
> If your connection is reliable and fast, you probably want to increase
> the checkpoint size from a gigabyte (the default) to, say, 16 or 64
> gigabytes: --checkpoint=64G
>
> I hope that helps. Since I think this isn't a real bug, I am closing
> it.Thanks for your answer. Would it make sense for obnam to default to a larger checksum size for a backup where both sides are local (as in not using sftp explicitly)? Ansgar
