topic Re: ntp issue on linux server, time goes ahead in Operating System - Linux

ntp issue on linux server, time goes ahead

AVV — Tue, 20 Jul 2010 19:45:49 GMT

Hi All,

I am using a BL 460c blade for a linux (rhel 5.3 64bit)host and an oracle dbase runs in. I am facing an issue with the timing now. If I am restarting the ntpd service, time sync with the server but slowly it goes ahead upto 10 min in two weeks which makes my database screwed. All other server connected to my time server is working well. I have given few outputs below. Please help me in this.

[root@prodsrvr log]# dig timesrv.dc

; <<>> DiG 9.3.4-P1 <<>> timesrv.dc
;; global options: printcmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 20578
;; flags: qr rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 0

;; QUESTION SECTION:
;timesrv.dc. IN A

;; AUTHORITY SECTION:
. 10759 IN SOA a.root-servers.net. nstld.verisign-grs.com. 2010072001 1800 900 604800 859

;; Query time: 0 msec
;; SERVER: 192.168.0.20#53(192.168.0.20)
;; WHEN: Tue Jul 20 23:38:40 2010
;; MSG SIZE rcvd: 108

[root@prodsrvr log]# nslookup timesrv.dc.com
Server: 192.168.0.20
Address: 192.168.0.20#53

Name: timesrv.dc.com
Address: 192.168.0.20

[root@prodsrvr log]#
[root@prodsrvr log]# ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
timesrv.dc .LOCL. 1 u 20 64 377 0.248 -268.90 136.550
[root@prodsrvr log]#

Re: ntp issue on linux server, time goes ahead

Tim Nelson — Tue, 20 Jul 2010 19:51:19 GMT

Is ntpd running ?

service ntpd status

Re: ntp issue on linux server, time goes ahead

AVV — Tue, 20 Jul 2010 20:08:12 GMT

Hi,

ntpd is running. I have confirmed that first.

#service ntpd status
ntpd (pid 11107) is running...

Re: ntp issue on linux server, time goes ahead

Wilfred Chau_1 — Tue, 20 Jul 2010 20:21:45 GMT

How come timesrv.dc.com resolves to an 192.168.0.0 IP ? Does it connect to an external NTP source?

Try change your ntp to use a public NTP server nearest to your machine from this url:
http://support.ntp.org/bin/view/Servers/NTPPoolServers

Re: ntp issue on linux server, time goes ahead

Matti_Kurkela — Tue, 20 Jul 2010 22:29:06 GMT

If there is no asterisk (*) in the first column in the "ntpq -p" output, the ntpd service has _not_ synchronized with a NTP server, so it cannot correct the system clock effectively.

Your offset value (estimated difference of the time value between your ntpd and the NTP server) is -268.90 ms: this is currently too much. For successful synchronization, the offset would have to be smaller than 128 ms (positive or negative).

The root cause for non-synchronization is probably the high jitter: if the round-trip delay in the communication between ntpd and the NTP server has more than about 100 ms of jitter, ntpd will drop out of sync so often that it will be almost useless. (The NTP documentation says the jitter value is an exponentially weighted root-mean-square average deviation from the average delay.)

In practice, the large jitter value means that the travel time of the NTP time message from the NTP server to your ntpd is not predictable enough. Jitter usually becomes high when you're trying to synchronize time over a very congested/overloaded network connection.

If you have other servers connected to the same switch as this BL460c blade, the congested connection might be the one between the blade and the switch. If possible, you should take a look at the switch port statistics and find out how busy the blade's network interfaces are.

If you have only one active network interface, consider setting up bonding to get more network bandwidth to your blade.

MK

Re: ntp issue on linux server, time goes ahead

bullz — Wed, 21 Jul 2010 06:55:27 GMT

I had the similar problem.

It's a VM?

If so, please login to your VC console and check the time at your VC console

It might help you.

Re: ntp issue on linux server, time goes ahead

AVV — Wed, 21 Jul 2010 09:57:34 GMT

This is not a VM but a physical blade server BL 460c

Re: ntp issue on linux server, time goes ahead

John Guster — Wed, 04 Aug 2010 17:07:03 GMT

what is the content of the file /etc/ntp.conf?

Re: ntp issue on linux server, time goes ahead

Ricky_HK — Wed, 11 Aug 2010 08:33:36 GMT

For me, I don't use NTP service.
Just add /usr/sbin/ntpdate -s to root crontab, let cron run it one or more times everyday... simple and easy.

Re: ntp issue on linux server, time goes ahead

Elmar P. Kolkman — Wed, 11 Aug 2010 10:30:52 GMT

You should try to setup a better NTP network. Have multiple servers (peers) use their clocks to provide the time to the rest of the network and let all other machines on your network use multiple NTP servers for time synchronisation, so that the best server is chosen.

Also have you own local clock in the ntp.conf (server 127.127.1.1).

Re: ntp issue on linux server, time goes ahead

Elmar P. Kolkman — Wed, 11 Aug 2010 10:43:18 GMT

One other thing: make sure a correct Proliant Support Pack is installed... It is needed to set the time in the hardware clock.

Re: ntp issue on linux server, time goes ahead

Bill McNAMARA_1 — Fri, 13 Aug 2010 03:51:21 GMT

check the ntpd -q output and the sysconfig options..

for reference:

[root@oclab50 ~]# ntpq -p
remote refid st t when poll reach delay offset jitter
==============================================================================
153.16.4.137 .INIT. 16 u - 64 0 0.000 0.000 0.000
ns1.your-site.c .INIT. 16 u - 64 0 0.000 0.000 0.000
c-68-51-42-35.h .INIT. 16 u - 64 0 0.000 0.000 0.002
LOCAL(0) .LOCL. 10 l - 64 0 0.000 0.000 0.002
[root@oclab50 ~]# cat /etc/sysconfig/ntpd
# Drop root to id 'ntp:ntp' by default.
OPTIONS="-x -u ntp:ntp -p /var/run/ntpd.pid"

# Set to 'yes' to sync hw clock after successful ntpdate
SYNC_HWCLOCK=no

# Additional options for ntpdate
NTPDATE_OPTIONS=""

Re: ntp issue on linux server, time goes ahead

AVV — Wed, 05 Jan 2011 06:46:07 GMT

Thanks