On Fri, Mar 25, 2011 at 11:31:19AM -0700, Chuck Swiger wrote: > Running ntpd is not easier than setting Xen's independent_wallclock = 0, or > VMware's vmx option tools.syncTime = true. As for best, well, I've yet to > see data where that happens: > > "6) I have yet to see an example where ntpd running in a DomU/guest kept > better time than using independent_wallclock = 0, tools.syncTime = true, etc > if the Dom0/host OS is sync'ed."
There is no independent_wallclock (or anything else that used to be in /proc/sys/xen) for Linux domU running mainstream kernels with paravirt-ops. It still exists on 2.6.18.8 with the legacy Xen patches, I believe, but that was a long long time ago. Interestingly, I have not noticed any NTP-related anomalies during periods of high load. CPU contention (identified by a nonzero steal%) is relatively rare everywhere except Slicehost, but even there, it does not seem to impact NTP's metrics. IO contention is more common, but (as you might expect) has no effect. > ...or a long thread on comp.protocols.time.ntp (starting from Message-id: > <568225d2-c7f0-4416-940b-9374da4d8...@glegroupsg2000goo.googlegroups.com>) > where a VM was seeing: It looks like the original poster is running a pre-pvops Linux kernel within Virtual PC running on a Windows system. That will fail majestically at any timing-critical task, like keeping time. There are a number of different virtualization technologies in use, and just because a few of them don't play well with ntpd doesn't mean that all virtualization is immediately bad. This was the case years ago, but not any more. Also, as requested, here's some ntpq -pcrv from a variety of VPSen: # Linode (Newark, 512 MB) # aka: http://www.pool.ntp.org/scores/97.107.134.213 rtucker@witte:~$ ssh framboise ntpq -pcrv assID=0 status=06f4 leap_none, sync_ntp, 15 events, event_peer/strat_chg, version="ntpd [email protected] Fri Dec 4 18:18:33 UTC 2009 (1)", processor="i686", system="Linux/2.6.32.16-linode28", leap=00, stratum=2, precision=-20, rootdelay=65.676, rootdispersion=23.570, peer=63002, refid=99.27.25.65, reftime=d1376b75.d71e4b43 Fri, Mar 25 2011 19:28:53.840, poll=10, clock=d1376d04.4ec15567 Fri, Mar 25 2011 19:35:32.307, state=4, offset=-0.142, frequency=4.648, jitter=1.172, noise=0.221, stability=0.018, tai=0 remote refid st t when poll reach delay offset jitter ============================================================================== -ntp3.tamu.edu 18.26.4.105 2 u 206 1024 377 53.796 -3.206 0.026 +tt155.ripe.Net .GPS. 1 u 145 1024 377 89.838 0.727 0.098 rhn-proxy01.dll 192.5.41.41 2 u 44d 1024 0 39.842 -0.325 0.000 -hennepin.ipv6.h 140.99.51.114 3 u 130 1024 377 49.925 -14.647 2.657 -ns1.anycast.cac 204.9.54.119 2 u 118 1024 377 19.193 0.867 0.004 +AC-NTP1.NET.CMU 128.10.254.7 2 u 200 1024 377 13.491 0.028 0.535 -morose.quex.org 66.187.233.4 2 u 243 1024 377 26.618 -4.730 0.136 *99-27-25-65.uvs .PPS. 1 u 400 1024 377 65.676 -1.059 0.088 # Slicehost (STL-A, 256 MB) rtucker@witte:~$ ssh <redacted> ntpq -pcrv assID=0 status=0644 leap_none, sync_ntp, 4 events, event_peer/strat_chg, version="ntpd [email protected] Fri Apr 9 03:13:54 UTC 2010 (1)", processor="x86_64", system="Linux/2.6.32.9-rscloud", leap=00, stratum=3, precision=-20, rootdelay=105.261, rootdispersion=76.019, peer=59966, refid=91.189.94.4, reftime=d13765fd.3e4a247b Fri, Mar 25 2011 19:05:33.243, poll=10, clock=d1376d11.e0ed5c88 Fri, Mar 25 2011 19:35:45.878, state=4, offset=-1.422, frequency=-41.286, jitter=0.065, noise=1.534, stability=0.079, tai=0 remote refid st t when poll reach delay offset jitter ============================================================================== *europium.canoni 193.79.237.14 2 u 788 1024 377 96.915 -1.422 1.271 # Rackspace Cloud (ORD, 512 MB) rtucker@witte:~$ ssh <redacted> ntpq -pcrv assID=0 status=0674 leap_none, sync_ntp, 7 events, event_peer/strat_chg, version="ntpd [email protected] Fri Apr 9 03:13:54 UTC 2010 (1)", processor="x86_64", system="Linux/2.6.35.4-rscloud", leap=00, stratum=3, precision=-20, rootdelay=54.861, rootdispersion=72.824, peer=42528, refid=69.36.227.90, reftime=d137693f.e8b4a991 Fri, Mar 25 2011 19:19:27.909, poll=10, clock=d1376d1e.c129a3aa Fri, Mar 25 2011 19:35:58.754, state=4, offset=-0.796, frequency=-24.309, jitter=4.690, noise=1.660, stability=0.114, tai=0 remote refid st t when poll reach delay offset jitter ============================================================================== +dosaku.ctipc.co 216.218.192.202 2 u 1023 1024 377 24.916 4.854 0.149 -vps1.cobryce.co 64.235.98.66 3 u 777 1024 377 58.442 -4.326 0.246 *name3.glorb.com 69.36.224.15 2 u 991 1024 377 50.604 -3.617 0.015 +lasvegas-nv-dat 67.18.187.111 3 u 738 1024 377 44.986 -1.490 11.508 # My desktop PC (HP with quad-core AMD Phenom II X4 830, not virtualized # :-) rtucker@witte:~$ ntpq -pcrv assID=0 status=0684 leap_none, sync_ntp, 8 events, event_peer/strat_chg, version="ntpd [email protected] Fri Apr 9 03:13:54 UTC 2010 (1)", processor="x86_64", system="Linux/2.6.32-30-generic", leap=00, stratum=3, precision=-20, rootdelay=33.886, rootdispersion=60.348, peer=12205, refid=38.229.71.1, reftime=d13767ce.04bfbe09 Fri, Mar 25 2011 15:13:18.018, poll=10, clock=d1376d27.4c957687 Fri, Mar 25 2011 15:36:07.299, state=4, offset=0.284, frequency=138.043, jitter=2.325, noise=2.388, stability=0.126, tai=0 remote refid st t when poll reach delay offset jitter ============================================================================== *clock.team-cymr 172.16.65.22 2 u 343 1024 377 32.467 -1.551 1.147 +mail.melancthon 138.39.23.13 2 u 33 1024 377 45.915 1.739 2.221 +206-217-140-200 184.106.219.19 3 u 85 1024 377 32.386 0.096 3.271 -- Ryan Tucker <[email protected]>
signature.asc
Description: Digital signature
_______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
