#!/bin/sh

# The file which will save the current position in the job
BREAKPOINT=/mydir/breakpoint
MAXPROCEDURE=3
#
# All the global variables should be declared here
#

save_break_point()
{
  echo $BP >$BREAKPOINT
}

procedure_1()
{
  # First part of the job
  echo procedure 1
}

procedure_2()
{
  # Second part of the job
  echo procedure 2
}

procedure_3()
{
  # Third part of the job
  echo procedure 3
}

#
# Initial state
[ -e $BREAKPOINT ] && BP=`cat $BREAKPOINT | sed 's/^\([0-9]*\).*$/\1/'`
[ "$BP" = "" ] && BP=1

# Process each function based on the breakpoint

while [ $BP -le $MAXPROCEDURE ]
do
  procedure_$BP
  save_break_point
  BP=`expr $BP + 1`
done
[ -e $BREAKPOINT ] && rm -f $BREAKPOINT

http://www.linuxquestions.org/questions/programming-9/bash-goto-384407/

you could divide your job into functions
Here, procedure_1 is the first part of the job, where there isn't restart, procedure_2 must be executed if procedure_1 was done and the control logic is at the end, working as an outer loop.

Unix Script control flow

Bash