Browse Source

Currently, after 4 failed constraint checks, we suspect the constraint

of being wrong, not the NTP responses, reset it and query it from all
the constraint servers all over again.  This is turned out to be a bit
aggressive because it could get triggered with just a few bad NTP
peers in a larger pool.  To avoid constant reconnections, scale the
error margin with the number of resolved NTP peers using peer_cnt * 4.
This way a single or a few outliers in a NTP pool cannot trigger
reconnecting to the constraint servers immediately.  More NTP peers,
less reason to mistrust the constraint.
Found by dtucker@
OK deraadt@
OPENBSD_5_8
reyk 9 years ago
parent
commit
aff57f6d18
1 changed files with 4 additions and 2 deletions
  1. +4
    -2
      src/usr.sbin/ntpd/constraint.c

+ 4
- 2
src/usr.sbin/ntpd/constraint.c View File

@ -1,4 +1,4 @@
/* $OpenBSD: constraint.c,v 1.9 2015/05/17 18:31:32 reyk Exp $ */
/* $OpenBSD: constraint.c,v 1.10 2015/05/18 14:19:23 reyk Exp $ */
/*
* Copyright (c) 2015 Reyk Floeter <reyk@openbsd.org>
@ -66,6 +66,7 @@ void *httpsdate_query(const char *, const char *, const char *,
char *tls_readline(struct tls *, size_t *, size_t *, struct timeval *);
extern u_int constraint_cnt;
extern u_int peer_cnt;
struct httpsdate {
char *tls_host;
@ -560,7 +561,8 @@ constraint_check(double val)
if (((val - constraint) > CONSTRAINT_MARGIN) ||
((constraint - val) > CONSTRAINT_MARGIN)) {
/* XXX get new constraint if too many errors happened */
if (conf->constraint_errors++ > CONSTRAINT_ERROR_MARGIN) {
if (conf->constraint_errors++ >
(CONSTRAINT_ERROR_MARGIN * peer_cnt)) {
constraint_reset();
}


Loading…
Cancel
Save