rsync and network failure

[Colin Law]

On 7 April 2016 at 09:53, Dave Howorth <dhoworth at mrc-lmb.cam.ac.uk> wrote:
>
>
> On 2016-04-07 09:21, Colin Law wrote:
>>
>> On 6 April 2016 at 14:57, Dave Howorth <dhoworth at mrc-lmb.cam.ac.uk> wrote:
>>>
>>>
>>>
>>> On 2016-04-06 14:43, Colin Law wrote:
>>>>
>>>>
>>>> I have a script that I run to keep a remote backup of a folder
>>>> heirarchy.  The script uses rsync with a command of the form
>>>>
>>>> rsync -H -azipc /path/to/backup user at my.remote.domain:destination_folder
>>>>
>>>> This works fine.  My question is, if the network fails part way
>>>> through, or the PC is shut down then should I just be able repeat the
>>>> command and rsync will transfer any remaining files that were not
>>>> successfully transferred the first time?
>>>
>>>
>>>
>>> I've never had any trouble. I've used dirvish for backups and downloads
>>> for
>>> many years, which uses rsync, and I've experienced various network
>>> problems.
>>> Dirvish just retries the rsync and normally the transfer completes the
>>> second time.
>>>
>>> Worst case on the rare occasion things somehow get completely mangled is
>>> just to delete the faulty image, but I usually do that after the
>>> following
>>> day's transfer has completed successfully.
>>
>>
>> You say you have never had any trouble, but then say that on rare
>> occasions things get mangled.  The two statements appear to be in
>> conflict.
>
>
> Sometimes it is not possible for rsync to recover. For example when the
> source server goes away permanently (because it has moved to a new address),
> or the DNS gets messed up, or the rsync version is changed at one end but
> not the other, or ... I'm sure there are more cases but I don't remember
> them all.

OK, understood.

>
>>  The reason I am asking is that this is intended to be an
>> automatic unattended operation (weekly via cron or something similar)
>> and I need to know that it won't get mangled, and
>
>
> I left it unattended for over a year at one point. No screams from the
> users.

Thanks

Colin