ucs 6248 HA cluster can not login

[hailin huang]

6248-A manager ip1.1.1.2

6248-B manager ip1.1.1.3

cluster VIP 1.1.1.1

cisco6248-A# show cluster extended-state
Cluster Id: 0xd59a70527a4111e2-0x8b74547feebbcd64

Start time: Sat Feb 23 12:44:03 2013
Last election time: Sat Feb 23 12:55:11 2013

A: UP, PRIMARY
B: UP, SUBORDINATE

A: memb state UP, lead state PRIMARY, mgmt services state: UP
B: memb state UP, lead state SUBORDINATE, mgmt services state: UP
   heartbeat state PRIMARY_OK

INTERNAL NETWORK INTERFACES:
eth1, UP
eth2, UP

HA READY
Detailed state of the device selected for HA storage:
Chassis 1, serial: FOX1618GW6D, state: active
Chassis 2, serial: FOX1625GAB0, state: active
Chassis 3, serial: FOX1613G1LV, state: active

MY PC IP 1.1.1.10 can ping all 1.1.1.1/2/3.and https all

but when i reload 6248-A ,can not ping 1.1.1.1,but can ping 1.1.1.3.

also can not https://1.1.1.1  (login error:failed login info:ucsm is not available on secondary node)

when 6248A reload ok,can ping 1.1.1.1/2/3,and https all

if fi6248 cluster is not ok?

[padramas]

Hello,

When you reload FI-A ( active primary ) , virtual IP is owned by FI-B, and should be become primary. Once it becomes primary, you should be able to access UCSM or SSH into FI via Virtual IP address ( 1.1.1.1 ).

1) How are these FI mgmt ports connected to upstream switch / L3 gateway ?

2) Are you able to SSH into FI-B with 1.1.1.1 ( when FI-A is reloading ) ?

3) Have you configured any mgmt interface monitoring policy ?

http://www.cisco.com/en/US/docs/unified_computing/ucs/sw/gui/config/guide/2.0/b_UCSM_GUI_Configuration_Guide_2_0_chapter_0101010.html#d75788e1551a1635

Padma

[hailin huang]

when FI-B becomes primary,(login error:failed login info:ucsm is not available on secondary nodeashould) unable to access UCSM or SSH into FI via Virtual IP address ( 1.1.1.1 ).

1.all FI mgmt ports connected to upstream switch ,and have a L3 gateway.

2.i able  to SSH into FI-B with 1.1.1.1 ( when FI-A is reloading ).

3.i have not configured any mgmt interface monitoring policy .

other place have same question.

[gdebroux]

Dear,

I'm having exactly the same issue on my Fabrics, running 2.04b.

Did anyone resolve this issue?

Thanks in advance and kind regards,

Gerald.

[sandevsingh]

Hi, is your upstream core Juniper?

[hailin huang]

noT JUNIPER .it is h3c.

[gdebroux]

We ran a new test with a linux host on the same subnet than the UCSM.

By looking at the ARP table on the linux host, we noticed that the VIP address changes within a minute.

So, it should be an upstream issue towards the subnet from where we access the UCSM.

Can you also run a test within the same subnet and verify if the VIP changes within a minute?

I believe we also have Junipers upstream. I'm investigating this now.

[sandevsingh]

We had experienced the same with Juniper, BUT when we upgraded to one of the latest JUNOS, it took less than a minute for the VIP`s arp to point to FI-B. 

[Robert Burns]

You might want to check that you have Juniper's equivelant of "PortFast" configured on the ports.  This might speed up the STP bring up process.

Regards,

Robert