rsync - Jun 2003 - Request for option for use on an unreliable connection

I have been using rsync for the last year or so. Generally I am very 
pleased with the performance and features, but I am forced to use a 
rather unreliable link to one of my mirrors. Consequentially rsync often 
bails out with a protocol error in the middle of a large session due to 
transient misbehaviour on the link. This appears to abort the whole 
transfer. One solution would appear to be an option to log the error 
then attempt the next file in the set. Is there a patch to allow this? 
Has anyone had similar problems?

TIA

David Lambert

On Wed, Jun 25, 2003 at 01:28:32PM -0500, David Lambert
wrote:
> I have been using rsync for the last year or so. Generally I am very 
> pleased with the performance and features, but I am forced to use a 
> rather unreliable link to one of my mirrors. Consequentially rsync often 
> bails out with a protocol error in the middle of a large session due to 
> transient misbehaviour on the link. This appears to abort the whole 
> transfer. One solution would appear to be an option to log the error 
> then attempt the next file in the set. Is there a patch to allow this? 
> Has anyone had similar problems?
If, after the file list phase, an error occurs on a file the
abort will only affect that file and those that follow.  All
previously processed files have been updated.  Many people
with unreliable links loop until success.  Each successive
run will get you a bit further along.

-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw@pegasys.ws

		Remember Cernan and Schmitt

jw schultz wrote:
>On Wed, Jun 25, 2003 at 01:28:32PM -0500, David Lambert wrote:
>  
>
>>I have been using rsync for the last year or so. Generally I am very 
>>pleased with the performance and features, but I am forced to use a 
>>rather unreliable link to one of my mirrors. Consequentially rsync often
>>bails out with a protocol error in the middle of a large session due to 
>>transient misbehaviour on the link. This appears to abort the whole 
>>transfer. One solution would appear to be an option to log the error 
>>then attempt the next file in the set. Is there a patch to allow this? 
>>Has anyone had similar problems?
>>    
>>
>
>If, after the file list phase, an error occurs on a file the
>abort will only affect that file and those that follow.  All
>previously processed files have been updated.  Many people
>with unreliable links loop until success.  Each successive
>run will get you a bit further along.
>  
>
Thanks for your reply. I have already tried what you suggest. My problem 
is that preceding files often get changed after being mirrored. Starting 
the whole process again gives "priority" to earlier files in the set. 
With one of my links this results in a virtually infinite loop. If rsync 
just moved to the next file, I think this would give a more even chance 
for all files in a set to update. Any other suggestions?

TIA

Dave.

I am reposting this request thread, as I did not see my reply posted, 
and have had no further replies.
>
>> I have been using rsync for the last year or so. Generally I am very 
>> pleased with the performance and features, but I am forced to use a 
>> rather unreliable link to one of my mirrors. Consequentially rsync 
>> often bails out with a protocol error in the middle of a large 
>> session due to transient misbehaviour on the link. This appears to 
>> abort the whole transfer. One solution would appear to be an option 
>> to log the error then attempt the next file in the set. Is there a 
>> patch to allow this? Has anyone had similar problems?
>>   
>
>
> If, after the file list phase, an error occurs on a file the
> abort will only affect that file and those that follow.  All
> previously processed files have been updated.  Many people
> with unreliable links loop until success.  Each successive
> run will get you a bit further along.
>  
>
Thanks for your reply. I have already tried what you suggest. My problem 
is that preceding files often get changed after being mirrored. Starting 
the whole process again gives "priority" to earlier files in the set. 
With one of my links this results in a virtually infinite loop. If rsync 
just moved to the next file, I think this would give a more even chance 
for all files in a set to update. Any other suggestions?

TIA

Dave.