SUMMARY date -a help.

Dear Managers,

Thanks to all who answered my "date -a" question.

Most of you suggested the usage of NTP which in most of the cases is the
right choice. There is lot of information one can find to configure and
enable the protocol with the respective benefits.

"date -a" is also an alternative to implement in emergency cases.
Unfortunately this one is poorly documented but once you have tested it the
man pages make sense. As I experienced, delaying the clock will take 5
hours to slowly delay the system in one hour (about one second per minute).
Advancing the clock will take less time to accomplish (I haven't had the
time to test it).

People, you are the best.

Here are the contributions:

Anthony Talltree <aad@verio.net>
I suggest running xntpd.

"Kevin A. Sindhu" <kevin.a.sindhu@Mail.AC>
The best solution is you run a internal time server (not open to any
other clients other than the ones in your network). Run the ntpd on one
of the machines and setup your solaris clients to sync with that ntp
server using ntpdate.

Or simply add, ntpdate tick.usask.ca into a bash script.

Eric Noriega <noriega@gwu.edu>
date -a doesn't permanently slow the clock, it causes the clock to
skew in a direction, until the specified number of seconds are made up.
Then the clock is restored, and will begin to exhibit it's natural
drift.  date -a is better then setting the date explicitly, because time
continues to look linear, instead of just jumping forward.

    If you need correct real time, use xntp.

Tim Chipman <chipman@ecopiabio.com>

date -a adjusts the time very gradually, IIRC, to avoid "time warp"
syndrome. (ie, wham! it is now 10 minutes earlier than it was a moment
ago! :-)

if you don't worry about time warp (nothing terribly time-sensitive
running).. you can probably force a synch against an external NTP time
source with "ntpdate ntp.server.ip.address" command

to keep servers all synched for their time, we're running NTP here - it
is pretty simple to setup and takes care of itself with no hassle.

IIRC, we ended up setting up one machine locally as the "local NTP time
source" - and this machine was the only one which synched NTP against
one of 4 possible external "second strata" NTP sources that were
moderately nearby.  A quick google search for "NTP stratum 2 time
source" will get you plenty of NTP servers available for (responsible,
moderate) use.

after we had one machine running locally as the "NTP time authority" we
simply configured all other machines to use this one machine as their
NTP authority.  If you are really worried, you can have 2 local auths
and configure local clients to use either of the 2 local NTP
authorities, depending on how mission-critical a few seconds of drift is
for your environment :-)

sample /etc/inet/ntp.conf files are shown below.

note that when ntp is running (starts automatically at startup on
systems where /etc/inet/ntp.conf exists, BTW) - you expect
/usr/lib/inet/xntpd daemon to be running, and would expect output approx
thus. from "ntpq -p" command.

local-ntp-slave # ntpq -p
     remote           refid      st t when poll reach   delay
offset    disp
==============================================================================

*local.ntp.master 199.212.17.15    3 u  361  68m  377     1.92
2.299    2.96
 LOCAL(0)         LOCAL(0)         9 l   60   64  377     0.00
0.000   10.03

Hope this helps,


Tim


LOCAL NTP MASTER - consults the outside world for NTP values:

server ntp1.cmc.ec.gc.ca prefer minpoll 8 maxpoll 12
server ntp2.cmc.ec.gc.ca minpoll 8 maxpoll 12
server time.chu.nrc.ca minpoll 8 maxpoll 12
server timelord.uregina.ca minpoll 8 maxpoll 12

# and apparently, we can always fall back to the local clock.
server 127.127.1.0

fudge 127.127.1.0 stratum 9

enable auth monitor
driftfile /var/ntp/ntp.drift
statsdir /var/ntp/ntpstats/
filegen peerstats file peerstats type day enable
filegen loopstats file loopstats type day enable
filegen clockstats file clockstats type day enable

All other Local NTP clients - consult only the local NTP master:
server 192.168.1.209 prefer minpoll 8 maxpoll 12
server 127.127.1.0
fudge 127.127.1.0 stratum 9

enable auth monitor
driftfile /var/ntp/ntp.drift
statsdir /var/ntp/ntpstats/
filegen peerstats file peerstats type day enable
filegen loopstats file loopstats type day enable
filegen clockstats file clockstats type day enable

--- End Of Message ---
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers