CentOS 5 :: Clock Time Jumps Into The Future (Xen + Ntpd)?
Nov 20, 2009
I'm having a problem with the time on one of my servers jumping forward into the future. We run serveral CentOS 5.4 servers running Xen, some up to date and some a little older. The one in question is running 2.6.18-164.el5xen, with xen-libs-3.0.3-94.el5_4.2 and xen-3.0.3-94.el5_4.2. On this server are several Xen virtual machines, also running CentOS 5.4 with kernel 2.6.18-164.el5xen. One in particular has problems with the clock keeping proper time.
We use NTP (ntpd) to sync time to a central server in the company. Our NTP set-up works fine, and we have problems on only a couple of servers out of many. The problem I see is that one one particular Xen virtual machine, the clock will suddenly jump forward into the future, usually by several minutes but once by more than an hour. This has obvious implications for software running on the machine. Software such as Oracle grid control agent will restart itself under the (incorrect) assumption that it has been hung for several minutes with no activity. Oracle database will cope gracefully, but applications that refer to the clock will be confused.
We detect these time jumps because we have Nagios checking the clock on each host against the centralised time server (Nagios's check_ntp plugin). Nagios will suddenly report a clock offset that is miles into the future. Following the time jump, ntpd on the host in question will re-sync the time. Ntpd keeps the time steady, ticking just a few milliseconds per second until real time catches up with the server. i.e. Nagios will report the time as being 10 minutes ahead, then a minute later the host will be 9 minutes ahead, and then a further minute later the host will be 8 minutes ahead of real time. The clock on the host stays running very slow until it eventually is correct.
I have a server running VMWare Server on it with 4 virtual machines running. I am trying to use NTPD to keep the time correct on all of the virtual machines. On my physical server, I have installed ntpd and this is it's config file:
Quote:# ntpd.conf tinker panic 0 restrict 127.0.0.1 restrict default kod nomodify notrap server 0.vmware.pool.ntp.org server 1.vmware.pool.ntp.org server 2.vmware.pool.ntp.org
I'm trying to use the ntpd -gq instead of ntpdate. The result is that clock not set. as below. However ntpdate is working ok and had set the clock correctly
This is the ntp.conf file: Code: [root@CentOS5 ~]# cat /etc/ntp.conf logfile /var/log/ntp.log driftfile /var/lib/ntp/drift server 0.centos.pool.ntp.org
This is the output of command: Code: [root@CentOS5 ~]# ntpd -dqg ntpd 4.2.2p1@1.1570-o Sat Dec 19 00:58:16 UTC 2009 (1) addto_syslog: precision = 1.000 usec create_sockets(123) addto_syslog: ntp_io: estimated max descriptors: 1024, initial socket boundary: 16 ..... clock_filter: n 6 off 42905.489022 del 0.390840 dsp 0.000132 jit 4.733917, age 12 16 addto_syslog: no reply; clock not set
Also attached is the trace file, from the trace file, everything seems ok, timestamps is sent and received:
When I shut down the machine everything is as it should be. Time set from the net, hwclock synced. When I boot again, time is set several hours into the future. This is an additive process, when I don't set the correct time it drifts with every boot even days into the future (until boot is denied, because files are too far in the future). I tried to get rid of that behaviour by doing:
Code: sntp -P no -r pool.ntp.org hwclock --systohc I renamed /etc/adjtime which looked like this:
22292.201093 1289670517 0.000000 1289670480 UTC
Now it looks as follows: 0.000000 1289733620 0.000000 1289733620 UTC
To no avail. What could cause that and how can I correct it?
I'm trying to sync my clock. I think I would like to do this by the "ntpd" daemon running. I think it should start up on boot, but it doesn't.
I have the following in my /var/log/syslog: Jul 18 23:44:14 neo ntpd[1780]: ntpd 4.2.6p2@1.2194-o Sun Oct 17 13:45:13 UTC 2010 (1) Jul 18 23:44:14 neo ntpd[1781]: proto: precision = 0.634 usec Jul 18 23:44:14 neo ntpd[1781]: unable to bind to wildcard address 0.0.0.0 - another process may be running - EXITING Jul 18 23:44:14 neo ntpdate[1194]: step time server 72.26.125.125 offset -0.505108 sec
Other posts say if ntpd and ntpdate run too close together, they cause one to think the other is locking a resource. It works to type "ntpdate pool.ntp.org" to manually update the clock. My Debian version is squeeze (6.0.2)
It works to start the ntpd daemon manually: ... sudo /etc/init.d/ntp start ... Starting NTP server: ntpd.
I have a F14 laptop being used as a music server. For some reason the clock time runs fast and slowly drifts forward about 5 minutes/day.
ntpd is running. The ntpd.conf is pretty much F14 out of the box and the same as my other F14 instances that do not exhibit this problem. If I restart ntpd, it syncs up the time, but then the time slips until the ntpd seems to give up.
Initially ntpd reports it is synched,but then over time it reports being unsynchronized.
Shortly after restart:
Code: ntpdc> sysinfo system peer: w1-wdc.ipv4.got-root.me.uk system peer mode: client leap indicator: 11 stratum: 4
I am trying to synchronize the time of my VM server with ntpd. I have the following configuration. And in the /etc/ntp.conf, I have the following line:
restrict default ignore restrict 127.0.0.1. server time1.server server time2.server
Whenever I have this line, the erver is not able to synchronize its time... So far as I understood, this line prevents other servers using this machine as a time server. And the second line says to allow localhost to use as time server. But why do I need to use its own time server when I have specified to use time1.server and time2.server ? ( firewall for tcp and udp ports 123 is open)
However, when I replace the first line of the configuration with the following line, it works. restrict default kod nomodify notrap noquery But with this, i am allowing other servers to use this server a ntp (which I wouldn't like to). Why this machine tries to use ntp server of its own (to snyc time) and why it is not working though i have the entry "restrict 127.0.0.1" ??
Just installed F15 using preupgrade (from F14) and now, every time I reboot, time jumps backwards 4 hours. If I reboot twice in a row without correcting the time, time will be off by 8 hours. Time zone is set correctly (Eastern Time). "Synchronize date and time over the network" is unchecked.
Since the difference between GMT and Eastern time is 4 hours, there appears to be some confusion over that but I can't find anything in the "Date & Time" app relative to that to try to play with to see if I can correct this. II do remember a similar issue many releases ago but I can't remember how it was dealt with.
I run a mostly untouched CentOS server, nothing is installed except a server app to store data. The server has been crashing lately - after a year of no problems, so I looked into the logs of it and noticed weird discrepancies in the time around the time it crashs. So I started "watch date >> datefile" and this is what happens:
Code: Sun Oct 17 09:52:45 CEST 2010 Sun Oct 17 09:52:47 CEST 2010 Sun Oct 17 09:52:49 CEST 2010
I am facing issue in ntp, it is getting killed frequently. see below error; # /etc/init.d/ntpd status ntpd dead but pid file exists After starting service it is works fine but again after sometimes it get in dead postition.
I have 9.04 and I recently "got" a 13.2GB 1080p HD movie. When I try to play it, it's choppy and mosaic like. The audio plays fine, but the video jumps and I can't get a clean picture except for maybe 15% of the time. I have to hold down the spacebar so it plays and pauses very quickly for the video to even proceed. I've updates VLC and Movie Player and I've downloaded many codecs and I can't seem to fix it.EDIT: I can play it in Movie Player without the pixel/mosaic-like effect, but it too is extremely choppy.
Recently upgraded to Centos 5.5 and was suprised to see that the distro still is running php 5.1.6.Anyone know when Centos RH plans on upgrading php in future distros?
I had cloned a centos 5.6 installation from virtualbox virtual machine to physical box. Everything work fine. However, the time showing in os using date command differs from bios time by roughly 4 hours. I am running ntp services which sync the time with another centos server on the network. It appears that some services are using virtual clock and some use physical clock. How do I get rid of virtual clock and only use physical clock?
I have a set of machines on a disconnected network. Periodically, one of the machines connects to the internet and synchronizes its time with a time server that is not known until the connection is established. (The machine queries a central command server for the address of the time server it should synchronize to.)
I then use a custom tool to do some calculations to call adjtimex() and adjust the clock so that it runs fairly accurately.
I know ntpd is supposed to be able to handle disconnected networks but I thought you had to preconfigure the servers in the configuration file.
My intent is to run ntpd on this machine (without configured "server"s) so that it can serve time to the internal network. (Periodic synchronization using ntpdate from the internal machine to the bridge machine.)
The problem: ntpd wants to fuss with the values I set using adjtimex(). I want it to quit thinking it needs to adjust the clock and just serve time to the internal network. (Maybe I have a GPS time source hooked directly to the machine!)
I have a Windows 2003 server that runs a Win32 time server which synchronizes with off site time sources. I have a CentOS 5.4 server which I just configured and installed NTP on. For some reason, I keep getting time discrepancies even after the server is (apparently) synching with my local NTP server on the Windows box.
Hwclock is set to UTC true/etc/localtime is properly symlinked to appropriate /usr/share/zoneinfo fileNTPD is setup as a service which loads on startupNTP.conf is configured to only retrieve from my local Windows 2003 time serverntpq -p seems to verify that the system is indeed synching up properlyThe problem is, whenever I run the 'date' command it is displaying my system time in (presumably) UTC. That is, even though my the CentOS box appears to be synching up to the local time, it's not translating it into my proper time zone. Consequently, the date is always at -6 hours from the actual local time. It was my understanding that the hwclock is fine to remain at UTC as ntpd should be synching with my local Windows 2003 time server periodically and updating it, and the system time should be getting translated to my localtime.
what I do, after rebooting the box the hwclock and system clock display the improper time.Maybe I just have a fundamental misunderstanding of how this is supposed to operate.Update: I'm not sure why, but after I commented out the other servers and added a fudge declaration for my internal server and rebooted, everything is working as it should and the time is correct.I'm still not really sure if it was adding the "fudge <hostname> stratum 2" declaration that corrected things, or commenting out the additional hosts, so I'll leave this open if anyone cares to comment or contribute.
i'm trying to use the ntpd -gq instead of ntpdate. The result is that clock not set. as below. However ntpdate is working ok and had set the clock correctly.this is the ntp.conf file:
[root@CentOS5 ~]# cat /etc/ntp.conf logfile /var/log/ntp.log driftfile /var/lib/ntp/drift server 0.centos.pool.ntp.org
this is the output of command: [root@CentOS5 ~]# ntpd -dqg ntpd 4.2.2p1@1.1570-o Sat Dec 19 00:58:16 UTC 2009 (1)[code]....
Also attached is the trace file, from the trace file, everything seems ok, timestamps is sent and received:
I am running a centOS 5.5 server and can't synchronize with time servers via ntpd. My server sits behind a firewall with only port 80 open and I suspect this is the issue. Do I need to open other ports to connect to the time servers at [URL]?
I cant get ntpd to synchronize with the ntp servers. I have a CentOS 5.3 32 bit Virtual Machine. I'm in USA on Pacific Standard Time. Here's all configuration and other important information for troubleshooting. Is there anything wrong with my config? btw, I used wireshark to look at the UDP packets - I dont get any responses from any of the servers back. I only see requests going out from my server, but no response. Is this a firewall Issue? If it is, then do I need to just enable UDP port 123 on the firewall?
When installing I perhaps mistakingly told Fedora it should use the BIOS clock and now it shows the wrong time, 1 hour ahead of my time-zone GMT+0. If I try and go to the preferences and set the time the 'Advanced Options' one of which I need are grayed out, I need the 'Use local time source' option to be unchecked. Could somebody tell me a workaround or the command-line commands to tell it not to do this anymore?
Trying to adjust my clock settings!I am running KDE 4.4.3 on Squeeze, on an 64bit laptop.(I used the AMD64 net install version)How can I fix the time settings so it shows 5:00pm instead of 17:00:00So far I have not been able to find a cure
When I boot into Debian (lenny) my clock is always 4 hours slow. Whether I set it manually, or set it to get it's time from the network, the next time I boot into debian, it's back to being 4 hours slow.
I have dual boot on my comp. Windows XP and Fedora 11 Now in both systems time zone is set to Belgrade ( which is my time zone), but when I setup clock in fedora to be, let's say 16.15h, then when I swich to windows it says time is 14.15h. When I setup in windows on 16.15h, and I swich to fedora, it says time is 18.15h. So I can't get accurate time on both systems in no way.
By default, Fedora 11 sets my clock to military time. For example it says 16:22. I would like it to show civilian time (or at least know how to do it) I logged into the clock settings and had to put in the root password, but couldn't find where you do this. If you scroll on the time for hours, it just goes from 0 to 23 and back, not to AM and PM like some others.
