openssh-unix-dev Digest, Vol 123, Issue 13

[Iain Morgan]

This seems to have helped. The unmbers look reasonable when the transfer
is resumed, but now there is similar strangeness when the transfer is
interrupted.

After the interrupt, a second status line now appears, which sometimes
reports unrealistically large or negative rates.

sftp> reget quuxfoo
Resuming /u1/imorgan/quuxfoo to quuxfoo
/u1/imorgan/quuxfoo                                                   80% 9949MB 109.0MB/s   00:21 ETA^Interrupt
/u1/imorgan/quuxfoo                                                   81% 9978MB -79437.-2KB/s   00:-2
sftp>

(This was using a 12GB testfile created with dd if=/dev/urandom.)

On Wed, Jul 31, 2013 at 16:30:37 -0500, Loganaden Velvindron wrote:
> Hi Iain,
> 
> Can you please try this patch ?
> 
> Index: progressmeter.c
> ===================================================================
> RCS file: /cvs/openssh/progressmeter.c,v
> retrieving revision 1.40
> diff -u -p -r1.40 progressmeter.c
> --- progressmeter.c	2 Jun 2013 13:46:24 -0000	1.40
> +++ progressmeter.c	31 Jul 2013 21:27:25 -0000
> @@ -147,7 +147,7 @@ refresh_progress_meter(void)
>  	if (elapsed != 0)
>  		cur_speed = (transferred / elapsed);
>  	else
> -		cur_speed = transferred;
> +		cur_speed = transferred - *counter;
> 
>  #define AGE_FACTOR 0.9
>  	if (bytes_per_second != 0) {
> 

-- 
Iain Morgan