USNO Washingon NTP servers down July 28-29, 2007

USNO Washingon NTP servers down July 28-29, 2007

Post by Rich Schmi » Sat, 21 Jul 2007 04:07:25


Due to scheduled maintenance upstream on the network, it is
anticipated that connectivity to all USNO Washington, DC NTP servers
will be down from 28 July 2007 1200 UT for a period of up to 24
hours. See http://www.yqcomputer.com/
as they become available.
Richard Schmidt
 
 
 

USNO Washingon NTP servers down July 28-29, 2007

Post by David L. M » Sat, 21 Jul 2007 06:01:37

Rich,

Holy Cow! Your Washington DC traffic baseline is about 5600 packets per
second, but once per hour it peaks over 10,000 packets per second. Where
are those varmits coming from? This suggests there are almost 5000
cronjobs running once per hour on the same minute, rather a dismal practice.

Dave

 
 
 

USNO Washingon NTP servers down July 28-29, 2007

Post by Wolfgang S » Sat, 21 Jul 2007 10:24:46


"David L. Mills" < XXXX@XXXXX.COM > writes:

Maybe stock ntpdate and SNTP cohorts could add a random number of
seconds to wait if it looks like they are being run near "even" times
(such as on the hour or perhaps half hour). Maybe the abusers will
pick up that code and have their load peaks diffused a bit.

It is amazing how much extra network traffic one sees at
round-numbered times. Cron makes it far too easy for folks to bunch
up requests at the same time.

-wolfgang
--
Wolfgang S. Rupprecht http://www.yqcomputer.com/
IPv6 on Fedora 7
http://www.yqcomputer.com/