How to erase a ACI Controller Config

[udo.konstantin]

Hi,

how is it possible to "erase" a Controller even I can't access to it.

We build our ACI fabric an configure the first controller, then the second and third. After login via browser into the first controller I notice the fabric is discovered an ready to configure. 

The other two controller are doesn't show as active. Because one of them seems to be in the initial setup wrong configured. 

Also it isn's possible to login into the other two controllers because we got a message "REST Endpoint user authorization is not initialized - Check Fabric Membership Status of this fabric node"

The initial setup for the second and third doesn't ask for a password. It should normally use the one which is configured on the first controller. 

Any idea to solve this? 

Thanks 

Udo 

[Tomas de Leon · Accepted solution]

Access the APICs via the CIMC, KVM.

Warning: You are accessing the ACI APIC Controllers!
rtp-f1-p1-apic3 login: rescue-user
Password:

**** The Password will be the password that you used in the "Setup" script or will be blank if never set.

rtp-f1-p1-apic3# whoami
admin

Then run the folloing commands:

acidiag touch clean
acidiag touch setup
acidiag reboot

Go thru the setup script again.

I hope this helps!

Cheers

T.

[nbastida90]

Hi Tomas,

I got the same problem here, but in my case I misscofigured the APIC, it has been configured as an standby node but It has to be configured as a active node for x3 cluster.

This error appears when I tryed to log in via GUI

"REST Endpoint user authorization datastore is not initialized - Check Fabric Membership Status of this fabric node"

I tryed to log into the apic with rescue-user and those password without any luck

- blank

- password

- "admin set password"

Any advice?

[udo.konstantin]

Hi Thomas,

thank you...that was the solution. 

We doesn't found the right password..;-) 

Regards

Udo 

[Tomas de Leon]

Udo,

It is hard to say what caused the issue without looking at the problem in a broken state & thru the logs.  If I had to guess, the most likely issue was something was fat fingered when going thru the setup script. Like the fabric name, node or pod id, infra-vlan, etc...

If APIC2 & APIC3 cannot communicate with APIC1 then you will not be able to login until they join the cluster successfully. 

But hard to say without seeing it in a broken state.  I am glad that it is working now.

Cheers!

T.