NTP time drift on SUP-7 based Switches

Ich Nafi · ‎04-13-2016

Hello everyone,

i stumbled upon the fact, that all our Sup-7-E based devices (4500X, 45xx-E, with different IOS versions) claim do be synchronized with our NTP servers, but all of them do have a time drift of 1-7 Minutes. Every other device (Nexus, Catalyst, various routers, you name it) that use the same NTP servers, are OK.

We are talking of about 50 Sup-7-E devices and over 1000 other devices which are OK. Is there anyone, who can give me a hint on this behavior?

Cheers

Stefan

Joseph W. Doherty · ‎04-13-2016

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Are you saying, NTP shows the 4500 is sync'ed, but if you show clock it's 1 to 7 minutes from the NTP server's time?

Or, are you saying if you show calendar, its time is several minutes off? If the latter, are you also using the command, NTP update-calendar?

Ich Nafi · ‎04-13-2016

Are you saying, NTP shows the 4500 is sync'ed, but if you show clock it's 1 to 7 minutes from the NTP server's time?

Exactly like this.

An example:

WS-C4500X-16:

sh ntp status
Clock is synchronized, stratum 2, reference is xxx.xxx.xxx.xxx
nominal freq is 250.0000 Hz, actual freq is 249.9990 Hz, precision is 2**10
ntp uptime is 2512750304 (1/100 of seconds), resolution is 4016
reference time is DAB8B2FD.58F434E1 (13:58:53.347 met Wed Apr 13 2016)
clock offset is -1.4793 msec, root delay is 0.94 msec
root dispersion is 66.26 msec, peer dispersion is 1.12 msec
loopfilter state is 'CTRL' (Normal Controlled Loop), drift is 0.000003772 s/s
system poll interval is 1024, last update was 4097 sec ago.

And here "show clock" on 3 different switches at the same time (2x 4500X-16, 1x C4506-E)

15:13:39.002 met Wed Apr 13 2016
15:14:04.676 MET Wed Apr 13 2016
15:14:52.590 met Wed Apr 13 2016

Show clock on a C2960S-48TS-L issued at the same moment (with the correct time)

15:14:47.995 met Wed Apr 13 2016

Joseph W. Doherty · ‎04-13-2016

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Interesting. I just manually did a fast switch (quarter second?) with a show clock on two different EX4500s, with sup7s, and got:

18:48:35.498 UTC Wed Apr 13 2016

18:48:48.943 UTC Wed Apr 13 2016

Ich Nafi · ‎04-13-2016

And was any of the times the correct one?

Btw.: I used SuperPutty to issue the show clock simultaneously to all 4 SSH sessions.

Joseph W. Doherty · ‎04-14-2016

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

That I don't know. The fact that they differ could be an indication of a similar issue as you describe.

Ich Nafi · ‎11-09-2016

Just to update this thread: I just realized, that the hardware-clock is running with the correct NTP time.

After executing a "clock read-calendar" the IOS clock shows the same time as the NTP. After a few days, IOS time drifts off again, while hardware clock is still correct.

Another Update: This behavior affects all of our SUP-7-E devices, with production date before 2014 (so serial number xxx17xxxxxx and before).

Ich Nafi · ‎11-17-2016

TAC confirmed, I hit the Bug CSCug75425.

Time to evaluate a new IOS Version....