Remove complex retry code and use ServerAliveInterval

Remove complex retry code and use ServerAliveInterval

Specifically pings 3 times after 60 secs each if no output from server.

BUG=chromium-os:10867
TEST=Ran it with Simple update test from test harness (image_to_live + run_remote_tests)

[sosa, 2011-01-14 17:52:51]

The only actual change (besides git revert's on my other two CL's) is adding the
-o ServerAliveInterval=60 to all SSH commands.

[petkov, 2011-01-14 18:19:23]

How does ServerKeepAlive actually help? You have a hanging connection that you
want to close after a certain timeout, right? This option will make the client
send more packets to try to keep the connection alive (so that it doesn't
timeout). I.e., this option is meant to keep the connection open, not to close
it (according to the documentation). Unless there's a side effect of ping
packets not getting responses but we don't really know if that's what's going to
happen.

The major issue is that we don't know why the connection is hanging.

http://codereview.chromium.org/6322002/diff/2001/remote_access.sh
File remote_access.sh (right):

http://codereview.chromium.org/6322002/diff/2001/remote_access.sh#newcode94
remote_access.sh:94: sleep .5
do you want to move this back into the "not yet rebooted" case?

[Mandeep Singh Baines, 2011-01-14 18:36:06]

petkov@chromium.org (petkov@chromium.org) wrote:
> How does ServerKeepAlive actually help? You have a hanging
> connection that you
> want to close after a certain timeout, right? This option will make
> the client
> send more packets to try to keep the connection alive (so that it doesn't
> timeout). I.e., this option is meant to keep the connection open,
> not to close
> it (according to the documentation). Unless there's a side effect of ping
> packets not getting responses but we don't really know if that's
> what's going to
> happen.
> 

From the man page:

"if the server becomes unresponsive ssh will disconnect after approximately
45 seconds."

This option will guarantee that at least the ssh connection doesn't get
hung.

> The major issue is that we don't know why the connection is hanging.
> 

Good point. We could be hung at some higher layer. But this will at least
rule out a bad ssh connection. It would be nice to get to the bottom
of what's going on. Maybe this CL fixes the issue, maybe not. But if it
doesn't, we rule out more stuff and get closer to understanding the problem.

> 
> 
> http://codereview.chromium.org/6322002/diff/2001/remote_access.sh
> File remote_access.sh (right):
> 
> http://codereview.chromium.org/6322002/diff/2001/remote_access.sh#newcode94
> remote_access.sh:94: sleep .5
> do you want to move this back into the "not yet rebooted" case?
> 
> http://codereview.chromium.org/6322002/

[petkov, 2011-01-14 18:42:33]

OK, LGTM. It might make sense to set ServerAliveCountMax explicitly though.

On 2011/01/14 18:36:06, Mandeep Singh Baines wrote:
> mailto:petkov@chromium.org (mailto:petkov@chromium.org) wrote:
> > How does ServerKeepAlive actually help? You have a hanging
> > connection that you
> > want to close after a certain timeout, right? This option will make
> > the client
> > send more packets to try to keep the connection alive (so that it doesn't
> > timeout). I.e., this option is meant to keep the connection open,
> > not to close
> > it (according to the documentation). Unless there's a side effect of ping
> > packets not getting responses but we don't really know if that's
> > what's going to
> > happen.
> > 
> 
> From the man page:
> 
> "if the server becomes unresponsive ssh will disconnect after approximately
> 45 seconds."
> 
> This option will guarantee that at least the ssh connection doesn't get
> hung.
> 
> > The major issue is that we don't know why the connection is hanging.
> > 
> 
> Good point. We could be hung at some higher layer. But this will at least
> rule out a bad ssh connection. It would be nice to get to the bottom
> of what's going on. Maybe this CL fixes the issue, maybe not. But if it
> doesn't, we rule out more stuff and get closer to understanding the problem.
> 
> > 
> > 
> > http://codereview.chromium.org/6322002/diff/2001/remote_access.sh
> > File remote_access.sh (right):
> > 
> > http://codereview.chromium.org/6322002/diff/2001/remote_access.sh#newcode94
> > remote_access.sh:94: sleep .5
> > do you want to move this back into the "not yet rebooted" case?
> > 
> > http://codereview.chromium.org/6322002/

[Mandeep Singh Baines, 2011-01-25 18:35:10]

ping

[sosa, 2011-01-25 18:56:23]

Sorry I committed it yesterday ... accidentally committed it with wrong issue
attached, but marking this as closed.