Nagios and NPCD delay system clock

[mksmr]

OpenBSD 6.0 GENERIC#2148 amd64
VMware vSphere 5.5
4 CPU
4GB RAM
VMware Tools running

Nagios 4.2.2 (built on this machine)
pnp4nagios from packages

Whenever I am starting Nagios, the system clock starts to drift excessively. As soon as Nagios is up, the clock deviation slowly rises to 1 to 3 seconds and eventually reaches a point where ntpd cannot correct it any longer. When I'm stopping Nagios, the deviation slowly decreases until it reaches its normal drift window of no more than 0.00x (normally 0.000x) seconds. It's even worse with NPCD, within seconds after it's start the clock begins to drift massively, ending up in being off by about 20 minutes in less than a day. Stopping NPCD slowly brings the system clock back on track.

EDIT: the clock always falls back, never ahead.

The problem is reproducable.

Basically, the machine in question is running Apache2, php70_fpm, Nagios 4.2.2, pnp4nagios-0.6.25p3, snmpd and ntpd - that's about it.

I have been running several Nagios instances on Linux machines and have never experienced this phenomenon. Any input on this would be greatly appreciated.

[dwhitfield]

Our OpenBSD experience is limited, so this may end up needing a community answer, but could you post your ntpd.conf? Thanks!

[mksmr]

Code: Select all

# $OpenBSD: ntpd.conf,v 1.14 2015/07/15 20:28:37 ajacoutot Exp $
server 192.168.2.10
constraints from "https://www.google.com"

Changing the server directive to other servers (i. e. pool.ntp.org) makes no difference.

[rkennedy]

Are you using vmware tools AND ntpd? In the past I've seen these two conflict each other and cause a drift.