help me understand keepalive..
steve at vellmont.com
Mon May 16 00:20:42 GMT 2005
Truly strange results. I tried --timeout=59 on the client side
(which is what my server config is set to), and got identical results
from before (timeout). I then tried a smaller value than what's
specified on the server config file, and it ran to completion.
Even specifying 58 seconds on the client side and it ran fine.
Whenever I specified a timeout on the client side of equal
to or greater than 59 seconds, I got a timeout at about 59 seconds
(including when I put in 120 seconds).
On Sun, May 15, 2005 at 04:15:46PM -0700, Wayne Davison wrote:
> On Sun, May 15, 2005 at 06:46:01AM -0500, Steve Sether wrote:
> > I applied this patch to 2.6.5pre1 and recompiled both sides
> > and am still getting the same error. My timeout is set to
> > 59 seconds.
> Hmm, it appears that your timeout is set in the server's config file,
> not via the client's command-line. Thus, the client cannot help to keep
> the connection alive. Since the client is the sender, my patch to make
> the sender send keep-alive messages during checksum gaps can have no
> effect without having a timeout in effect. You should try running the
> command again with --timeout=59 as a command-line arg.
> This does raise an interesting issue that the daemon protocol should be
> extended to share any config-set timeout value with the client so that
> both sides know about the restriction, and can contribute to keeping the
> connection alive.
More information about the rsync