aLIGO LHO Logbook

X1 SEI

david.barker@LIGO.ORG - posted 13:11, Thursday 14 October 2010 - last comment - 16:31, Friday 15 October 2010(211)

ntp client now working

found the problem which was preventing ntp client from running
on the sei frontend. There was an entry in the /etc/ntp/step-tickers
file (was a caltech address) so I had to add the GC NTP server's
address to this file. Seems that if this file is empty, ntp would
also work.

I kept the ntpdate crontab running on the workstation (every minute)
and that machine is time shifting a lot to keep up with the front
end (shifts of 0.1 second every minute are common). Not sure
what's up with that, I have now turned off the crontab and
ntpd is running sync'ing with the GC server also.

Comments related to this report

james.batch@LIGO.ORG - 10:51, Friday 15 October 2010 (212)

Link

This morning (Oct. 15) the time on the workstation differed by greater than 10 seconds from the seiteststand.  Looked at /var/log/messages, found that no ntpd synchronization had occurred since Oct. 11.  Stopped the ntpd on workstation, edited the ntp.conf file to add a 'restrict' line for the ntp server, sync'd the time with ntpdate, and restarted the ntpd.  Both workstation and seiteststand are now using the same ntp server, workstation is no longer using seiteststand as a server.  

Looking at /var/log/messages on seiteststand, there are a number of 'time reset' lines with large values.  This needs to be monitored and investigated.

james.batch@LIGO.ORG - 16:31, Friday 15 October 2010 (213)

Link

After monitoring the clock on a different workstation (of the same type as seiteststand), it appears the clocks are not very accurate.  A Supermicro 1U computer on our DAQ DTS test stand has gained 24, 28,and 19 seconds per hour over the last three hours.