Re: OA, blades and interconnects all on the same VLAN (without separate management network) (509 Views)
Reply
Occasional Visitor
Vartan Narinian
Posts: 4
Registered: ‎06-10-2000
Message 1 of 9 (921 Views)

OA, blades and interconnects all on the same VLAN (without separate management network)

Hi all,

Did a search but couldn't find anything similar...

We have a setup with a c7000 with two Cisco c3020,
each having an uplink (of type access, i.e. non-trunked)
to an infrastructure switch.

The OAs have their own separate connections to the infra switches.

All the iLOs, OAs, blades and EBIPA have addresses from
the same network, i.e. there is no separate management network
(yes, yes, I know this is not good practice...)

So the Cisco 3020 management interface (Fa0) gets an IP address
on the same VLAN as the blade servers.

Now the problem: From the blades in the same chassis (using bonding under
RHEL) you can only ping one of the 3020s, but not the other. But from an
external network (beyond the infra switches) you can ping both 3020s.

I'm sure it's something to do with the 3020 management interface Fa0
(which goes through the OA connection) being on the same VLAN as
the blades (which go via the 3020 uplinks).

Is there a way to fix this without altering the setup? (this would involve
a lot of bureaucracy).
Trusted Contributor
Diego Castelli
Posts: 269
Registered: ‎12-21-2005
Message 2 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

in this sit. can be an internal link between the cisco's.

try to issue:
show interface | include line|\/sec


and have a look at the ports that you can map to servers in OA.

Is there (apart from fa0) any difference?

i suspect an automatic STP...

Diego C.

MS MCSA Server 2003

HP Accredited Integration Specialist
Trusted Contributor
Diego Castelli
Posts: 269
Registered: ‎12-21-2005
Message 3 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

have a look also:
http://www.cisco.com/en/US/tech/tk389/tk621/technologies_tech_note09186a0080136673.shtml
Diego C.

MS MCSA Server 2003

HP Accredited Integration Specialist
Occasional Visitor
Vartan Narinian
Posts: 4
Registered: ‎06-10-2000
Message 4 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

I take your point about STP, but the problem
is that the Cisco 3020 won't give any info
for Fa0

Plus it works from outside networks.

And doesn't STP ensure that every single
network segment is accessible?
Trusted Contributor
Diego Castelli
Posts: 269
Registered: ‎12-21-2005
Message 5 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

Yeah, but could be that there's a link that's NOT automatically switched off, creating a loop.

What do you think would happen if you remove the link between the second switch and the infra switches?
Do you think that all blades mapped only to the second (unpingable) switch would be isolated?

Another thing:
Did you setup NIC teaming on the blades?
If so, could be that blades are going out on the first NIC (connected to the first switch) and refuses to reach the other switch cause of some confusion about MAC addresses...

Keep in mind that i'm doing suppositions here...

Sorry i can't help you more, but i don't have DIRECT experience with CISCO 3020 switches inside enclosures.

Here i ask for someone else shading light about "internal" connections.

One solution could be to separate Management VLAN and to configure your preferred solution for having the mgmt interfaces/IPs
Diego C.

MS MCSA Server 2003

HP Accredited Integration Specialist
Occasional Visitor
Vartan Narinian
Posts: 4
Registered: ‎06-10-2000
Message 6 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

I don't think there's a loop. But I made
another observation today.

You cannot ping switch 2 from switch 1
and vice versa.

It must be something to do with them using the
Fa0 interface. (All the arp entries are on
Fa0).

I updated the software on the switches today.
Before the upgrade, the blades could ping
sw2. I upgraded sw2 first. When it came back,
I upgraded sw1. The blades can now ping sw1
but not sw2.

Either switch can ping either OA.
Trusted Contributor
Diego Castelli
Posts: 269
Registered: ‎12-21-2005
Message 7 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

What about arp entries relative to the mgmt IP for switch 1 in switch 2 arp tables and vice versa?
are there static arp entries?

So you cannot ping switch 1 neither from the ping diagnostic utility of switch 2?
Really strange.
Diego C.

MS MCSA Server 2003

HP Accredited Integration Specialist
Occasional Visitor
Vartan Narinian
Posts: 4
Registered: ‎06-10-2000
Message 8 of 9 (921 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

That is correct. The arp entries in the
switches show as incomplete.

Even if I add arp entries manually in the
switches, they still cannot ping each other.

We have other enclosures at the same customer
and they all behave similarly, so it's not
a one-off.
Visitor
JordyL
Posts: 2
Registered: ‎05-15-2013
Message 9 of 9 (509 Views)

Re: OA, blades and interconnects all on the same VLAN (without separate management network)

Same issue.. Did it ever get resolved
The opinions expressed above are the personal opinions of the authors, not of HP. By using this site, you accept the Terms of Use and Rules of Participation.