ds2k5
Contributor
Contributor

ESXi 4.1 - ntp sync problems with 3 IBM eserver m326

Hello,

we have 3x IBM eserver m326 with ESXi 4.1 (build 381591).

and we have 1x HP Proliant DL320 G6.

All 3 IBM Servers have time sync problems if i open the vSphere Clinet and go to Confiuration -> Timeconfiguration

it is RED. If i set the TIME correct it runs a couple of hours, but drift then.

my ntp.conf is identical on all 4 servers.

restrict default kod nomodify notrap nopeer
restrict 127.0.0.1
server ptbtime1.ptb.de
driftfile /etc/ntp.drift

I have traced with: tcpdump-uw -c 5 -n -i vmk0 host 192.53.103.108 and port 123

( http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=100509... )

(vSphere Console say NTP is NOT OK)

~ # tcpdump-uw -c 5 -n -i vmk0 host 192.53.103.108 and port 123
tcpdump-uw: WARNING: SIOCGIFINDEX: Invalid argument
tcpdump-uw: verbose output suppressed, use -v or -vv for full protocol decode
listening on vmk0, link-type EN10MB (Ethernet), capture size 96 bytes
05:42:56.136924 IP 10.40.42.54.123 > 192.53.103.108.123: NTPv4, Client, length 48
05:42:56.202880 IP 192.53.103.108.123 > 10.40.42.54.123: NTPv4, Server, length 48
05:47:04.117932 IP 10.40.42.54.123 > 192.53.103.108.123: NTPv4, Client, length 48
05:47:04.182255 IP 192.53.103.108.123 > 10.40.42.54.123: NTPv4, Server, length 48
05:51:14.189726 IP 10.40.42.54.123 > 192.53.103.108.123: NTPv4, Client, length 48

(here i have set the time before start tcpdump-uw)

~ # tcpdump-uw -c 10 -n -i vmk0 host 192.53.103.108 and port 123
tcpdump-uw: WARNING: SIOCGIFINDEX: Invalid argument
tcpdump-uw: verbose output suppressed, use -v or -vv for full protocol decode
listening on vmk0, link-type EN10MB (Ethernet), capture size 96 bytes
08:50:50.970834 IP 10.40.42.41.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:50:51.038764 IP 192.53.103.108.123 > 10.40.42.41.123: NTPv4, Server, length 48
08:51:54.965547 IP 10.40.42.41.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:51:55.032657 IP 192.53.103.108.123 > 10.40.42.41.123: NTPv4, Server, length 48
08:52:58.959937 IP 10.40.42.41.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:52:59.026679 IP 192.53.103.108.123 > 10.40.42.41.123: NTPv4, Server, length 48
08:54:02.954947 IP 10.40.42.41.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:54:03.020612 IP 192.53.103.108.123 > 10.40.42.41.123: NTPv4, Server, length 48
08:55:06.949667 IP 10.40.42.41.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:55:07.016171 IP 192.53.103.108.123 > 10.40.42.41.123: NTPv4, Server, length 48

here from the HP Proliant (vSphere Console say NTP is OK)

~ # tcpdump-uw -c 5 -n -i vmk0 host 192.53.103.108 and port 123
tcpdump-uw: WARNING: SIOCGIFINDEX: Invalid argument
tcpdump-uw: verbose output suppressed, use -v or -vv for full protocol decode
listening on vmk0, link-type EN10MB (Ethernet), capture size 96 bytes
08:39:24.088458 IP 10.40.42.47.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:39:24.153098 IP 192.53.103.108.123 > 10.40.42.47.123: NTPv4, Server, length 48
08:56:29.088689 IP 10.40.42.47.123 > 192.53.103.108.123: NTPv4, Client, length 48
08:56:29.158597 IP 192.53.103.108.123 > 10.40.42.47.123: NTPv4, Server, length 48

I see there is a difference beetween the servers in the intervall, why ?

The Time in the BIOS is set "back" of 2 hours behind UTC

what did i wrong ?

sorry about my poor english

0 Kudos
3 Replies
ds2k5
Contributor
Contributor

here are the esxcfg-info output of one IBM an the HP Server

0 Kudos
AndreTheGiant
Immortal
Immortal

Have you tried this KB?

http://kb.vmware.com/kb/1005092

Andre | http://about.me/amauro | http://vinfrastructure.it/ | @Andrea_Mauro
0 Kudos
ds2k5
Contributor
Contributor

yes, from this KB i know the tcpdump-uw

0 Kudos