[Nagiosplug-devel] check_ntp misunderstandings?

[Gerhard Lausser]

Hi Sean

> perhaps we could throw in a new cmdline switch to have the 
> jitter check against the local ntpd instead of the remote one?
Such a pure-local switch would be helpful. It could also save
an extra service which checks if a ntp daemon is running on the
local host.
What comes to my mind is like this:

... in main
case 'l':      /* -l check jitter of the local ntpd /*
  localntp=1;
  break;
case 's':
  strictlocal=1;  /* -s be strict. critical if no local ntpd /*
  break;

... in do_jitter
double jitter_request(const char *host, int *status){
  if (localntp) {
    strcpy(server_address, "127.0.0.1");
  }
  my_udp_connect(server_address, 123, &conn);
  if (localntp && strictlocal && !conn) {
    DBG("no local ntp daemon running")
    return -2.0;
  }

... in main
   if(do_jitter) {
    ...
		} else if(jitter == -1.0 && result == STATE_OK){
			/* -1 indicates that we couldn't calculate the
jitter
			 * Only overrides STATE_OK from the offset */
			result = STATE_UNKNOWN;
		} else if(jitter == -2.0 && result == STATE_OK) {
                  /* -2 indicates there is no local ntp daeomon */
                  /* plus strictlocal is set */
                  result = STATE_CRITICAL;

... at the end of main
    if (do_jitter) {
      if (jitter == -2.0) {
        printf(", no local ntp daemon running");
      } else {
        printf(", jitter=%f", jitter);
      }

Checking for a running daemon would be very helpful, because it is a
platform-independent replacement for "check_proc -C [ntpd|xntpd|...]"

Greetings from Munich,
Gerhard