NTP - suggestion

[Jeeshan]

Hi All

I have some HP-UX servers, Running in 11iv1,v2 & v3. There is also a cluster environment in 2 9000 servers. But i'm facing the date problem on those servers. I'm gonna deploy the NTP server and synchronize all machines from that NTP server.
My point is that,
1.Is it necessary to sycnchronize the NTP server for external stratum?
2.is it necessary to synchronize my NTP server from internet time server?
3.Is there be any synchronization problem in cluster nodes?
4. Any downtime needed for the cluster nodes? or NTP server?

Please suggest.

[Anshumali · Accepted solution]

1. no. suggested=Yes
2. No. You can use local clock. Suggested ( Depends on your requirement and level of accuracy needed)
3. you need to ensure the time is sync'd between your cluster.
4. NTP client= May be if you are running DB and time is ahead of standard. This will require the downtime other wise no.

[skt_skt]

Here is our setup.

At client side.

adedd13n:root [/root] cat /etc/ntp.conf
#
#
# eXtended Network Time Protocol configuration file - ntp.conf
#
# Configuration file for all systems outside the stratum-3 peer group
#
# 11/30/95 Bj Miller Initial Coding (412)553-3126
#
#
# 05/12/2003 By Dan Copeland, Vijay Nallapati & Shankar Basa -- Change ticket #C0032030
# Replaced OLD stratum-2 servers (hpeas01, bkmstr01 and acapr01)

# The NTP peer group...
#
server server1.com version 3
server server2.com version 3
server server3.com version 3

#
# Allow server to sync to its own clock in case of network outage.
# Clock is configured as a stratum-9 clock

server 127.127.1.0
fudge 127.127.1.0 stratum 10 # show poor quality
driftfile /etc/ntp.drift

At server side(server1.com)

adedd13n:root [/root] ssh tnedd02p "cat /etc/ntp.conf"
#
#
# eXtended Network Time Protocol configuration file - ntp.conf
#
# Configuration file for all systems outside the stratum-3 peer group
#
# 11/30/95 Bj Miller Initial Coding (412)553-3126
#
#

# The NTP peer group...
#

server 0.us.pool.ntp.org
server 1.us.pool.ntp.org

peer server2.com
peer server3.com


# Allow server to sync to its own clock in case of network outage.
# Clock is configured as a stratum-9 clock

server 127.127.1.1
fudge 127.127.1.1 stratum 5
#fudge 127.127.1.1 stratum 10 # show poor quality
driftfile /etc/ntp.drift
server 127.127.1.1

[Jeeshan]

Thanks Anshumali

But i want to know for cluster environment do i need to synchronize both servers at a time? or first down the cluster and then up one node and synchronize?

[skt_skt]

if they are PROD nodes then i would recommend to shutdown the cluster/packages on the nodes and sync the time and bring the cluster up.

But you can test this w/o halting the packages on a DEV nodes. But you need to bring down DBs/appl and do the ntp sync bring up db/appl to pick the new time.. MCSG cluster is NOT that sensitive to time as compared to the CRS clusters.

[Ajay Gupta]

I don't think time need to be synchroinzed on all servers to run cluster but you should do that in order to keep consistancy accross all production servers.

[Jeeshan]

solved!!!