[ntp:questions] help - local ntp server not serving

Dave Hart davehart at gmail.com
Thu Mar 3 16:34:20 UTC 2011

On Thu, Mar 3, 2011 at 14:59 UTC, <hymie at lactose.homelinux.net> wrote:
> transmit: at 7> mode 3 len 48
> receive: at 7<- mode 4 len 48
> packet: flash header 1420

 * Define flasher bits (tests 1 through 11 in packet procedure)
 * These reveal the state at the last grumble from the peer and are
 * most handy for diagnosing problems, even if not strictly a state
 * variable in the spec. These are recorded in the peer structure.
 * Packet errors
#define TEST1		0X0001	/* duplicate packet */
#define TEST2		0x0002	/* bogus packet */
#define TEST3		0x0004	/* protocol unsynchronized */
#define TEST4		0x0008	/* access denied */
#define TEST5		0x0010	/* bad authentication */
#define TEST6		0x0020	/* bad synch or stratum */
#define TEST7		0x0040	/* bad header */
#define TEST8		0x0080  /* bad autokey */
#define TEST9		0x0100	/* bad crypto */
#define	PKT_TEST_MASK	(TEST1 | TEST2 | TEST3 | TEST4 | TEST5 |\
			TEST6 | TEST7 | TEST8 | TEST9)
 * Peer errors
#define TEST10		0x0200	/* peer bad synch or stratum */
#define	TEST11		0x0400	/* peer distance exceeded */
#define TEST12		0x0800	/* peer synchronization loop */
#define TEST13		0x1000	/* peer unreacable */
#define	PEER_TEST_MASK	(TEST10 | TEST11 | TEST12 | TEST13)

peer unreachable + peer distance exceeded + bad synch or stratum.  If
you run ntpd on the client normally as a daemon, and this problem
continues, you can use ntpq -classoc to get the association ID for
your server, then ntpq -c "rv ASSOCID" to fetch the association's
variables.  Look for dispersion, showing the estimated maximum error
along the NTP path.  This is also referred to as root distance or root
dispersion in some cases.  Apparently the sum of the source NTP's
server's own root distance plus the component from your hop to it
exceed the default "tos maxdist" threshold, which probably indicates a
problem with the server ntpd, or that it hasn't been running long

Dave Hart

More information about the questions mailing list