Remove complex retry code and use ServerAliveInterval

remote_access.sh

Index: remote_access.sh
diff --git a/remote_access.sh b/remote_access.sh
index 45531f15526aa4f08327e5732f08b3bb0e0b1155..7cbab7ec3347ef8d714ec752136bdbbff9577b55 100644
--- a/remote_access.sh
+++ b/remote_access.sh
@@ -16,7 +16,8 @@ DEFINE_integer ssh_port 22 \
 # Copies $1 to $2 on remote host
 function remote_cp_to() {
   REMOTE_OUT=$(scp -P ${FLAGS_ssh_port} -o StrictHostKeyChecking=no -o \
-    UserKnownHostsFile=$TMP_KNOWN_HOSTS $1 root@$FLAGS_remote:$2)
+    ServerAliveInterval=60 -o UserKnownHostsFile=$TMP_KNOWN_HOSTS $1 \
+    root@$FLAGS_remote:$2)
   return ${PIPESTATUS[0]}
 }
 
@@ -24,13 +25,14 @@ function remote_cp_to() {
 # $2.  Directory paths in $1 are collapsed into $2.
 function remote_rsync_from() {
   rsync -e "ssh -p ${FLAGS_ssh_port} -o StrictHostKeyChecking=no -o \
-            UserKnownHostsFile=$TMP_KNOWN_HOSTS" --no-R \
-    --files-from=$1 root@${FLAGS_remote}:/ $2
+            ServerAliveInterval=60 -o UserKnownHostsFile=$TMP_KNOWN_HOSTS" \
+            --no-R --files-from=$1 root@${FLAGS_remote}:/ $2
 }
 
 function remote_sh() {
   REMOTE_OUT=$(ssh -p ${FLAGS_ssh_port} -o StrictHostKeyChecking=no -o \
-    UserKnownHostsFile=$TMP_KNOWN_HOSTS root@$FLAGS_remote "$@")
+    ServerAliveInterval=60 -o UserKnownHostsFile=$TMP_KNOWN_HOSTS \
+    root@$FLAGS_remote "$@")
   return ${PIPESTATUS[0]}
 }
 
@@ -68,29 +70,12 @@ function learn_board() {
   info "Target reports board is ${FLAGS_board}"
 }
 
-# Checks to see if pid $1 is running.
-function is_pid_running() {
-  ps -p ${1} 2>&1 > /dev/null
-}
-
-# Wait function given an additional timeout argument.
-# $1 - pid to wait on.
-# $2 - timeout to wait for.
-function wait_with_timeout() {
-  local pid=$1
-  local timeout=$2
-  local -r TIMEOUT_INC=1
-  local current_timeout=0
-  while is_pid_running ${pid} && [ ${current_timeout} -lt ${timeout} ]; do
-    sleep ${TIMEOUT_INC}
-    current_timeout=$((current_timeout + TIMEOUT_INC))
-  done
-  ! is_pid_running ${pid}
-}
+function remote_reboot {
+  info "Rebooting."
+  remote_sh "touch /tmp/awaiting_reboot; reboot"
+  local output_file
+  output_file="${TMP}/output"
 
-# Checks to see if a machine has rebooted using the presence of a tmp file.
-function check_if_rebooted() {
-  local output_file="${TMP}/output"
   while true; do
     REMOTE_OUT=""
     # This may fail while the machine is down so generate output and a
@@ -101,23 +86,12 @@ function check_if_rebooted() {
     if grep -q "0" "${output_file}"; then
       if grep -q "1" "${output_file}"; then
         info "Not yet rebooted"
-        sleep .5
       else
         info "Rebooted and responding"
         break
       fi
     fi
-  done
-}
-
-function remote_reboot() {
-  info "Rebooting."
-  remote_sh "touch /tmp/awaiting_reboot; reboot"
-  while true; do
-    check_if_rebooted &
-    local pid=$!
-    wait_with_timeout ${pid} 30 && break
-    ! kill -9 ${pid} 2> /dev/null
+    sleep .5

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

do you want to move this back into the "not yet rebooted" case?

   done
 }