Add more robust ssh ping to be run on reboot for updates.

Add more robust ssh ping to be run on reboot for updates.

Upon investigation it seems there's a race condition when ssh
starts up the first time.  This adds a much more robust way to
ping a machine when it's up then the previous method.

Change-Id: I092744b259c169975d1c4f283f01a556fce24723

BUG=chromium-os:10867
TEST=Ran it 30 times with image_to_live --noupdate --nostateful_update.

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=6058752

[petkov, 2011-01-12 18:38:52]

mostly optional nits with one real bug, i think

http://codereview.chromium.org/6115009/diff/1/remote_access.sh
File remote_access.sh (right):

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode49
remote_access.sh:49: local current_timeout=0
this site shows an interesting way to implement a timeout that doesn't require
waiting for the full timeout yet doesn't do a loop:
http://redflo.de/tiki-index.php?page=Bash+script+with+timeout+function

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode50
remote_access.sh:50: while ps -p ${ssh_pid} > /dev/null && \
no need for trailing \ I think

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode51
remote_access.sh:51: [ ${current_timeout} -gt ${timeout} ]; do
current_timeout is 0, timeout is 5 -- this evaluates to false, no? so the loop
will never iterate?

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode53
remote_access.sh:53: echo "blah"
blah? :)

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode54
remote_access.sh:54: current_timeout=$((current_timeout+timeout_inc))
add spaces around +

http://codereview.chromium.org/6115009/diff/1/remote_access.sh#newcode59
remote_access.sh:59: else
s/else/fi/, or better:

kill ... || break

are you sure kill SIGTERM won't hang? should you just kill -9 instead?

[sosa, 2011-01-12 21:53:58]

PTAL

Didn't move to common library yet.  Maybe in another CL?

Also I think I cleaned this up a bit.  The retries are really unnecessary since
remote_reboot already does it.

[petkov, 2011-01-12 21:59:20]

Wait but no kill anymore?

[sosa, 2011-01-12 22:04:48]

Re-added kill

! kill to return inverse status ... i.e. we wanna return 1 if the pid still
exists after waiting.

[petkov, 2011-01-12 22:40:31]

LGTM

http://codereview.chromium.org/6115009/diff/9001/remote_access.sh
File remote_access.sh (right):

http://codereview.chromium.org/6115009/diff/9001/remote_access.sh#newcode48
remote_access.sh:48: local timeout_inc=1
maybe

readonly TIMEOUT_INC=1