SIP SERVER: band aid monitoring

oils_sip_mon.rb

# BAND AID SIP SERVER MONITOR

class SipServerMonitor

  def initialize(opts = {})
    @sip = {
      start: '/openils/bin/oils_ctl.sh -d /openils/var/run -s /openils/conf/oils_sip.xml -a start_sip',
      stop: '/openils/bin/oils_ctl.sh -d /openils/var/run -s /openils/conf/oils_sip.xml -a stop_sip',
      pid_file: '/openils/var/run/oils_sip.pid',
      pid: nil,
      name: "SIPServer",
      warn: 30,
      log: '/var/log/syslog',
      email: nil,
    }.merge! opts

    @sip[:hostname] = `hostname`
  end

  def check_pid(restart = false)
    pid = `cat #{@sip[:pid_file]}`
    unless pid.empty?
      @sip[:pid] = pid.to_i
    else
      report("OILS NO PID", "PID NOT FOUND IN #{@sip[:pid_file]}")
      # restart_server if restart
    end
  end

  def check_no_connections(restart = false)
    connections = `pgrep -f #{@sip[:name]} | wc -l`
    unless connections.empty?
      connections = connections.to_i
      if connections > @sip[:warn]
        report("SIP WARN PROCESSES", "SIP WARN PROCESSES EXCEEDED: warn=#{@sip[:warn]}: current=#{connections.to_s}")
        # restart_server if restart
      end
    else
      report("SIP NO PROCESSES", "NO SIP SERVER PROCESSES WERE FOUND")
      # restart_server if restart
    end
  end  

  def check_pid_match(restart = false)
    raise "MUST CHECK PID FIRST" unless @sip[:pid]
    ppid = `pgrep -f #{@sip[:name]} -P 1`
    unless ppid.empty?
      ppid = ppid.to_i
      if ppid != @sip[:pid]
        report("OILS PID AND SIP PID MISMATCH", "OILS PID AND SIP SERVER PROCID DO NOT MATCH")
        # restart_server if restart
      end
    else
      report("SIP NO PARENT PROCESS", "SIP SERVER PARENT NOT FOUND")
      # restart_server if restart
    end
  end

  def grep_log(search)
    if system( "grep #{search} #{@sip[:log]} > /dev/null" )
      report("FOUND #{search}", "FOUND #{search} IN #{@sip[:log]}")
    end
  end

  def report(subject, body)
    subject += " -- #{@sip[:hostname]}"
    system( "echo #{body} | mail -s #{subject} #{@sip[:email]}" ) if @sip[:email]
  end

  def restart_server
    system( @sip[:stop] )
    system( "pkill -f #{@sip[:name]}" )
    system( "rm #{@sip[:pid_file]}" )
    system( @sip[:start] )
  end

end

begin
  ssm = SipServerMonitor.new( { email: 'some.address@somewhere.com' } )
  ssm.check_pid # check oils_sip.pid
  ssm.check_no_connections # check there are processes, but not too many
  ssm.check_pid_match # check there is a parent and that it matches oils_sip.pid
  ssm.grep_log 'handle_patron_status' # expected ... error, misconfigured workstation possible
rescue Exception => ex
  puts ex.message
end

__END__

This filled a hole for a short while. But really use Monit (or similar): http://mmonit.com/monit/