Thanks.
When pinging either the hard-coded address or the DHCP assigned address,
the ping fails anytime we unplug the ethernet cable to that nic.
We think that the HMC has opened up a tunnel in linux which is acting as
a proxy for <someone/some process>. At first we thought maybe the the
laptops in the z10, but the ping does not fail if we disconnect the
cable going to the z10 switch.
I wish I could get into the real linux and doing some displays.
Tony Thigpen
Bigendian Smalls wrote on 11/21/2015 11:54 AM:
It occurs to me Tony, there could be a multitude things here. But one involves
routes and routing tables. Presuming the two networks on your multi-homed HMC
don't overlap (different subnets) I'd wonder how your routes in that box are
set up (default and otherwise).
#3 concerns me a little that you have the same Mac arp resolution to 2
different IPs. While not impossible there could be some issues with this. Have
you captured the traffic from the host that is suspect - and rebooted, see if
it is actually dhcp requesting an address? It sounds like there could possibly
be another box with a MAC address the same as your HMC. That's one idea. or a
process on it requesting a dhcp address.
If it were me I'd take two traffic captures, one that gets all the traffic from
the HMC you suspect. and one from the box handing out IPs. Do a reboot of the
HMC if you can, and see if the traffic coincides with expected behavior.
That's where I'd start.
Chad
On Nov 20, 2015, at 23:03, Tony Thigpen <[email protected]> wrote:
Background: HMC software version 2.11.1 connected to a z10.
The HMC is connected to two networks. The first is a small private network with
just the laptops in the z10 and the HMC. No other HMCs or other CPUs. The
second network is a local network with several items on it, including other
HMCs. This network is behind a VPN firewall and is used to access the web
services on the HMC.
Both interfaces have hard-coded IP addresses.
Today we noticed something strange. We had a DHCP address assigned to a box we
did not know about. Except for 3 workstations on the network, all other boxes
have hard-coded IP addresses. In the DHCP assigned addresses table, the box had
provided a name of BMC_DHCP.
After a bunch of testing, we isolated the assigned address to the HMC, but the
address does not show up in any of the HMC panels that show IP addresses.
Items:
1) When we ping the HMC using the hard-coded address, the response is under
3.0ms.
2) When we ping the DHCP assigned address, the response is 10x longer, in the
30.0ms range.
3) The arp tables on another PC show both addresses having the same nic address.
4) The HMC can ping it's own hard-coded address, but it can not ping the DHCP
assigned address.
5) We have other CPUs and other HMCs. None of the others are doing the same
thing. (z900 though z10).
6) This HMC has the latest software version of all the HMCs.
7) nMap (port mapping tool) says that there are no ports open at this DHCP
assigned address.
Thoughts on what is happening?
Anybody else seeing the same thing?
--
Tony Thigpen
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
