NTP Differences

pdexter86 · **Joined:** Mon Sep 14, 2015 10:55 am **Posts:** 8

Hi we have a 2 node 7200 with a difference in time between the nodes and the NTP server.

0 2016-05-11 15:38:40 BST (Europe/London)
1 2016-05-11 15:35:58 BST (Europe/London)

at the time I ran this the NTP was at 15:36 BST

Can anyone advise whether this is something I can resolve myself or will it need logging with HP?

Thanks

RitonLaBevue · **Joined:** Fri Jun 27, 2014 2:01 am **Posts:** 390

you can solve it yourself of course

use setdate

cali · **Posted:** Thu May 12, 2016 3:11 am

I see this behavior on nearly every 3PAR, that even if NTP is set, the Time diverges on tne Nodes.

koopa79 · **Joined:** Mon Apr 14, 2014 9:07 am **Posts:** 189

Cali,

So when you've seen this behavior before, what have you done to remedy the situation?

cali · **Posted:** Fri May 13, 2016 7:00 am

using setdate
again and again

ailean · **Joined:** Wed Nov 09, 2011 12:01 pm **Posts:** 392

I only started seeing this on recent 8400 systems, F/T/V seemed okay in the past.

setdate works for a while (days

) but I found the nodes still drifted off again. You can try changing the primary node for management which moves which node holds the active IP and runs the NTP requests.

That helped me a bit, in that less nodes were out of sync but I then had to reboot the remaining drifters which seemed to sort it.

Have since had 3.2.2eMU2 update installed, of course restarts the nodes and just checking I can see some drift again (up to 30 sec).

Support didn't seem too bothered about it and didn't really help at all except saying use setdate.

RitonLaBevue · **Joined:** Fri Jun 27, 2014 2:01 am **Posts:** 390

Years ago (2008 to 2011) , when i was a customer with 2 T400 i scheduled a daily batch to manually set the date on both arrays.

So that's not really a recent fact.

Josh26 · **Joined:** Thu Oct 24, 2013 6:50 pm **Posts:** 185

cali wrote:

I see this behavior on nearly every 3PAR, that even if NTP is set, the Time diverges on tne Nodes.

You know I ran into this yesterday, and before having a chance to Google it, I see this statement.

Interestingly, it causes a checkhealth to produce a big alert.

ailean · **Joined:** Wed Nov 09, 2011 12:01 pm **Posts:** 392

I do find that interesting, as I mentioned I've not seen this until recent 8400 systems.
F400 & T800 ran for over 5 years in sync and never ran setdate.
2 x V400 systems, 1-3 years and still in sync, never used setdate.
The V400s are to the second currently but I've seen maybe a drift of 1 sec between nodes when checking frequently (only since noticing the issue on the 8400s :roll:

).

ailean · **Joined:** Wed Nov 09, 2011 12:01 pm **Posts:** 392

Glad to see there maybe a patch to help with this now, P27.

Quote:

This patch includes the following modifications to 3PAR OS 3.2.2 MU2:
• Improves the reliability with which non-network master controller nodes sync to the network master controller node's NTP time.

HPE Storage Users Group

NTP Differences

Who is online