[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference smurf::buildhelp

Title:USG buildhelp questions/answers
Moderator:SMURF::FILTER
Created:Mon Apr 26 1993
Last Modified:Mon Jan 20 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2763
Total number of notes:5802

145.0. "ntpd using Bellevue, WA and Manalpan NJ for a time base!?" by SMURF::FILTER (Automatic Posting Software - mail to flume::puck) Tue Jun 15 1993 14:24

Date Of Receipt: 	15-JUN-1993 13:20:03.66
From: 	ABYSS::werme "Eric Werme USSG"
To: 	abyss::odehelp
CC: 	abyss::bstevens, abyss::netperf
Subj: 	ntpd using Bellevue, WA and Manalpan NJ for a time base!?

After noticing alingo was polling those sites for ntp data, I found this
in the ode setup script:

    if [ -f  /usr/sde/${def_prod_dir}/lib/templates/${domain}/settime.rcmgr.${os} ]
    then
	.  /usr/sde/${def_prod_dir}/lib/templates/${domain}/settime.rcmgr.${os}
    else
	/usr/sbin/rcmgr set ROUTED "yes"   
	/usr/sbin/rcmgr set XNTPD_CONF "YES"
        /usr/sbin/rcmgr set NFSLOCKING 1
	/usr/sbin/rcmgr set XNTP_SERV1 "16.64.0.246"   # ode.zso
	/usr/sbin/rcmgr set XNTP_SERV2 "16.140.64.9"   # merge.zk3
	/usr/sbin/rcmgr set XNTP_SERV3 "16.142.80.2"   # alphabits.aosg.gsf
	/usr/sbin/rcmgr set XNTP_SERV4 "16.69.224.139" # odie.unx
    fi
...
cat > /etc/ntp.conf <<!EOF
driftfile /etc/ntp.drift
peer 16.64.0.246 version 1 	# ode.zso
peer 16.140.64.9 version 1 	# merge.zk3
peer 16.142.80.2 version 1 	# alphabits
peer 16.69.224.139 version 1 	# odie.unx
!EOF

Not only are zso and unx useless to me as ntp servers thanks to net
latencies, this has overridden the defaults I used to get to a server
on my subnet.

ntp appears to send a message to *each* server once a minute.  If everyone
ran the ode installation script, we'd be exchanging a steady stream of
junk with both unx and zso.  The link to zso even goes out on a
commercial net that looks pretty congested already:

alingo 13% traceroute ode.zso.dec.com
traceroute to ode.zso.dec.com (16.64.0.246), 30 hops max, 40 byte packets
 1  r1-red (16.140.64.1)  2 ms  4 ms  2 ms
 2  spitbrook_w.zko.dec.com (16.33.0.5)  2 ms  3 ms  2 ms
 3  wrl-zko-pp.bb.dec.com (16.50.0.6)  622 ms  609 ms  508 ms
 4  easy-pa-gw1.pa.dec.com (16.1.224.100)  498 ms  554 ms  470 ms
 5  pa-ds1-gw1.pa.dec.com (16.1.240.34)  461 ms  685 ms  328 ms
 6  zs-pa.p-p.dec.com (16.10.16.36)  409 ms  447 ms  352 ms
 7  ode.zso.dec.com (16.64.0.246)  360 ms  480 ms  219 ms

alingo 12% traceroute odie.unx.dec.com
traceroute to odie.unx.dec.com (16.69.224.139), 30 hops max, 40 byte packets
 1  r1-red (16.140.64.1)  8 ms  3 ms  3 ms
 2  fastlink.zko.dec.com (16.33.0.2)  4 ms  3 ms  4 ms
 3  unxvlink.unx.dec.com (16.69.224.135)  15 ms  16 ms  16 ms
 4  odie.unx.dec.com (16.69.224.139)  16 ms  15 ms  19 ms

Also, there's extra bind traffic with unx because the install script
doesn't add odie.unx.dec.com to /etc/hosts.  (I search my local file
before asking bind.)

I'd recommend changing the installation script to not muck with an
existing ntp configuration, and then only adding hosts within the
same domain.

T.RTitleUserPersonal
Name
DateLines
145.1Re: ntpd using Bellevue, WA and Manalpan NJ for a time base!?SMURF::FILTERAutomatic Posting Software - mail to flume::puckTue Jun 15 1993 14:3179
Date Of Receipt: 	15-JUN-1993 13:06:39.39
From: 	QUARRY::myrdal "Gregory Myrdal USG  15-Jun-1993 1250"
To: 	[email protected] (Eric Werme USSG)
CC: 	[email protected], [email protected], [email protected]
Subj: 	Re: ntpd using Bellevue, WA and Manalpan NJ for a time base!?

I talked with Josh about this a few weeks back.  He was going to look into
fixing the install scripts.  Josh, any news on this front?  Thanks,

-- Greg

 *** Responding to your message dated Tue, 15 Jun 1993 11:52:28 -0400 :

>After noticing alingo was polling those sites for ntp data, I found this
>in the ode setup script:
>
>    if [ -f  /usr/sde/${def_prod_dir}/lib/templates/${domain}/settime.rcmgr.${
>os} ]
>    then
>	.  /usr/sde/${def_prod_dir}/lib/templates/${domain}/settime.rcmgr.${os}
>    else
>	/usr/sbin/rcmgr set ROUTED "yes"   
>	/usr/sbin/rcmgr set XNTPD_CONF "YES"
>        /usr/sbin/rcmgr set NFSLOCKING 1
>	/usr/sbin/rcmgr set XNTP_SERV1 "16.64.0.246"   # ode.zso
>	/usr/sbin/rcmgr set XNTP_SERV2 "16.140.64.9"   # merge.zk3
>	/usr/sbin/rcmgr set XNTP_SERV3 "16.142.80.2"   # alphabits.aosg.gsf
>	/usr/sbin/rcmgr set XNTP_SERV4 "16.69.224.139" # odie.unx
>    fi
>...
>cat > /etc/ntp.conf <<!EOF
>driftfile /etc/ntp.drift
>peer 16.64.0.246 version 1 	# ode.zso
>peer 16.140.64.9 version 1 	# merge.zk3
>peer 16.142.80.2 version 1 	# alphabits
>peer 16.69.224.139 version 1 	# odie.unx
>!EOF
>
>Not only are zso and unx useless to me as ntp servers thanks to net
>latencies, this has overridden the defaults I used to get to a server
>on my subnet.
>
>ntp appears to send a message to *each* server once a minute.  If everyone
>ran the ode installation script, we'd be exchanging a steady stream of
>junk with both unx and zso.  The link to zso even goes out on a
>commercial net that looks pretty congested already:
>
>alingo 13% traceroute ode.zso.dec.com
>traceroute to ode.zso.dec.com (16.64.0.246), 30 hops max, 40 byte packets
> 1  r1-red (16.140.64.1)  2 ms  4 ms  2 ms
> 2  spitbrook_w.zko.dec.com (16.33.0.5)  2 ms  3 ms  2 ms
> 3  wrl-zko-pp.bb.dec.com (16.50.0.6)  622 ms  609 ms  508 ms
> 4  easy-pa-gw1.pa.dec.com (16.1.224.100)  498 ms  554 ms  470 ms
> 5  pa-ds1-gw1.pa.dec.com (16.1.240.34)  461 ms  685 ms  328 ms
> 6  zs-pa.p-p.dec.com (16.10.16.36)  409 ms  447 ms  352 ms
> 7  ode.zso.dec.com (16.64.0.246)  360 ms  480 ms  219 ms
>
>alingo 12% traceroute odie.unx.dec.com
>traceroute to odie.unx.dec.com (16.69.224.139), 30 hops max, 40 byte packets
> 1  r1-red (16.140.64.1)  8 ms  3 ms  3 ms
> 2  fastlink.zko.dec.com (16.33.0.2)  4 ms  3 ms  4 ms
> 3  unxvlink.unx.dec.com (16.69.224.135)  15 ms  16 ms  16 ms
> 4  odie.unx.dec.com (16.69.224.139)  16 ms  15 ms  19 ms
>
>Also, there's extra bind traffic with unx because the install script
>doesn't add odie.unx.dec.com to /etc/hosts.  (I search my local file
>before asking bind.)
>
>I'd recommend changing the installation script to not muck with an
>existing ntp configuration, and then only adding hosts within the
>same domain.

---------------------------------------------------------------------
Gregory P. Myrdal                   TCP/IP: [email protected]
M/S ZKO3-3/V08                      UUCP:   decvax!myrdal
Digital Equipment Corporation       DECnet: WASTED::MYRDAL
110 Spit Brook Road                 Voice:  1-603-881-0304
Nashua, New Hampshire 03062-2698    FAX:    1-603-881-2257

145.2Re: ntpd using Bellevue, WA and Manalpan NJ for a time base!?SMURF::FILTERAutomatic Posting Software - mail to flume::puckTue Jun 15 1993 15:3618
Date Of Receipt: 	15-JUN-1993 14:05:14.06
From: 	WASTED::fran "Fran Fadden USG  15-Jun-1993 1405"
To: 	[email protected] (Eric Werme USSG)
CC: 	[email protected], [email protected], [email protected]
Subj: 	Re: ntpd using Bellevue, WA and Manalpan NJ for a time base!?

It would also be nice, if odesetup is being fixed anyway, for it not to
assume that routed is the correct router to use.  Something like the
following would do...

	if (x`/usr/sbin/rcmgr get ROUTED` != xyes && x`/usr/sbin/rcmgr get GATED` != xyes)
	then
		usr/sbin/rcmgr set ROUTED "yes"
	fi

					Thanks,
					Fran