[ntp:questions] Accept high root dispersion from Windows NTP server
Lucas van Braam van Vloten
lucas2 at dds.nl
Fri Jul 10 14:29:37 UTC 2015
Thanks for your reply.
Unfortunately, introducing a physical server in this environment just
for the NTP function is not an option.
The environment is actually a client's development and test environment
which runs entirely on VMs in some supplier's cloud.
I think that the time within the environment is reasonable accurate. My
Domain Controllers synchronize themselves against a different time
source (which is not accessible from my linux servers, by the way). But
my main concern is that time is identical on all servers within the
environment. It is of lesser importance whether the source of the time
is 100% reliable.
-----Original Message-----From: Cube Central <cubecentral at gmail.com>
To: 'Lucas van Braam van Vloten' <lucas2 at dds.nl>
Subject: RE: [ntp:questions] Accept high root dispersion from Windows
Date: Fri, 10 Jul 2015 08:09:08 -0600
The first, and maybe obvious or straightforward answer that pops into my
head is to reorganize this network service and get it off the Virtual
Machines. NTP just doesn't do well in VMs. I would create an independent
stand-alone ntp server and then point the windows servers and Linux
servers/clients to that. It would solve the problem right away without
needing to fool with how ntp works.
[mailto:questions-bounces+cubecentral=gmail.com at lists.ntp.org] On Behalf Of
Lucas van Braam van Vloten
Sent: Friday, 10 July, 2015 07:09
To: questions at lists.ntp.org
Subject: [ntp:questions] Accept high root dispersion from Windows NTP server
I am running ~30 Linux servers (RHEL6) which synchronize their time against
the NTP server on two Windows 2008 Domain Controllers.
My Linux clients can see the time servers, but association is rejected.
I believe that this is due to a very high root dispersion on these servers,
namely around 10s.
I have talked to the Windows administrator, and according to him this high
root dispersion has something to do with the server running on a VM
(VMWare). He says that it does not make the server a less reliable time
source, and that it cannot be solved on the Windows server. "We just have to
live with it" (his words)
However, my Linux NTP clients keep rejecting the association.
Is there a way in the NTP client configuration to ignore the root dispersion
value of the server in the decision whether to associate?
Or, alternatively, can I adjust the threshold for what is accepted, for
example to 15 seconds or so?
This is the output on my Linux machine:
ind assid status conf reach auth condition last_event cnt
1 61931 9024 yes yes none reject reachable 2
2 61932 9024 yes yes none reject reachable 2
ntpq> rv 61931
associd=61931 status=9024 conf, reach, sel_reject, 2 events, reachable,
srcadr=puhtdc0001.piaddot.prorail.nl, srcport=123, dstadr=172.24.75.21,
dstport=123, leap=00, stratum=2, precision=-6, rootdelay=31.250,
reftime=d94a40a7.f1dcccf5 Fri, Jul 10 2015 15:01:27.944,
rec=d94a4115.dcbd0077 Fri, Jul 10 2015 15:03:17.862, reach=377, unreach=0,
hmode=3, pmode=4, hpoll=6, ppoll=6, headway=56,
flash=400 peer_dist, keyid=0, offset=16.847, delay=0.662, dispersion=16.516,
filtdelay= 0.66 0.72 0.45 0.72 0.41 0.56 0.63
filtoffset= 16.85 28.33 19.69 27.05 31.19 22.19 22.18
filtdisp= 15.63 16.62 17.61 18.58 19.54 20.55 21.52
ntpq> rv 61932
associd=61932 status=9024 conf, reach, sel_reject, 2 events, reachable,
srcadr=puhtdc0002.piaddot.prorail.nl, srcport=123, dstadr=172.24.75.21,
dstport=123, leap=00, stratum=3, precision=-6, rootdelay=62.500,
rootdisp=11128.265, refid=172.24.75.30, reftime=d94a404a.0c65965c Fri, Jul
10 2015 14:59:54.048,
rec=d94a4114.de3c9281 Fri, Jul 10 2015 15:03:16.868, reach=377, unreach=0,
hmode=3, pmode=4, hpoll=6, ppoll=6, headway=54,
flash=400 peer_dist, keyid=0, offset=12.680, delay=0.733, dispersion=16.513,
filtdelay= 0.73 0.80 0.67 0.63 0.66 1.17 7.34
filtoffset= 12.68 17.04 14.66 13.59 19.40 19.83 28.60
filtdisp= 15.63 16.62 17.59 18.55 19.56 20.55 21.54
Any help would be greatly appreciated!
questions mailing list
questions at lists.ntp.org
More information about the questions