cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
3110
Views
0
Helpful
9
Replies

NTP DNS Lookup - Fails at Boot and then removed.

Rob Craig
Level 1
Level 1

I am hoping there is a quick solution to my query that I am not seeing in my own research.

IE-2000-8TC-B, devices running code 15.0(2)EA1

Configuration uses DNS for NTP and the domain name and 2 name-servers configured:

ntp server <DOMAINNAME.com>
ip domain-name <NAME.com>
ip name-server X.X.X.X
ip name-server X.X.X.Y

After a reload NTP attempts to resolve via DNS, I suspect before the uplink interfaces are fully up, therefore,  the name-servers are not reachable, the resolution fails and the configuration is removed from the running-config and does not attempt to retry connecting.

I can connect and enter the command ntp server <DOMAINNAME.com> and it resolves correctly and everything is fine.

I considered KRON jobs and other methods but it seems like I am over analysing something that potentially has a quick fix which I have missed.

Anyone encountered this or have a configuration suggestion?

Thanks in advance

9 Replies 9

Philip D'Ath
VIP Alumni
VIP Alumni

I'm not familiar with that platform, but the general syntax for IOS devices running 15.x code to do recurring DNS lookups (as opposed to a single lookup) for NTP is:

ntp server ip 1.pool.ntp.org
ntp server ip 0.pool.ntp.org

Hi Phil, thanks for responding.

If they both fail (due to startup and a data path not being available) do you think this will loop so they would try again? Fear I would be in the same predicament. That being no NTP synchronised with the relevant NTP configuration present in the startup but not the running configuration file. When using a static IP address the issue doesn't occur and the line makes it from the startup into the running with NTP synchronised.

On my kit - yes they keep looping.

The "ip" command causes a DNS lookup everytime it is used, do you can have NTP servers on dynamic IP addresses and it still works.

Thanks , will check with the ip in and see we how we go. 

Hi Philip,

I tried using the "ip" keyword, both with our internal NTP and the two you referenced. Issue persisted. I have attached some captures showing the commands in the file structures for: Running-Config, Startup-Config and Config.Text. The device is then reloaded and all of the NTP DNS entries are removed from all of the configurations. Strange to say the least.

Attempted a couple of things including.

NTP Source VLAN1.

NTP Source <uplink interface>

Clear NTP Drift

Different platform, different code: IE-2000-4TS-G-B on 15.0(1r)EY1

Nothing makes a difference, NTP information is always persistent if I use a straight IP.

I noticed that in the logging buffer from start up: NTP Core (INFO): keys initilized. Occurs before the interfaces are up, but unsure yet if this is related.

Any other thoughts? Thanks in advance.

I think it is highly likely you have run into a software bug.  Is their newer firmware for these switches?

I'll try a later version of code on my test box.. can't see any registered BUG-IDs.

Thanks for your time Philip. Much appreciated. Will let you know if I reach an outcome

Did you ever get a resolution on this. It appears I have the same issue on our 3850 switches running version 03.06.04.E software and on version 15.2(2)E4 on our 2960's. Both IOS are the recommended versions from Cisco.

Sorry Kevin, we had no joy and reverted to an IP implementation.

Good luck