Re-Install Hyperflex Cluster

[IC2317]

Hi,

I am looking for steps to re-install Hyperflex cluster from scratch. It does not have anything running on it and due to change in design we need to change the replication factor. As we cannot change the RF after installation, so I guess we need to re-install it again.

There is clean up disk partitions option in Cisco Intersight, However not sure if that that would be enough for the reinstallation.

Can someone please suggest.

[Kirk J]

Greetings.

The full doc is at  https://community.cisco.com/t5/unified-computing-system/hyperflex-customer-cleanup-guides-for-fi-and-edge/ta-p/3896668?attachment-id=174876  , but not all those steps are always necessary.

Here are the core steps needed:

• Move all workload guestVMs off HX cluster, or delete them.
• Delete HX datastores from HX connect
• from a storage controller SSH connection, run stcli cluster info | grep vCenterClusterId:
  • note the 'domain' name and number, as you'll need that for 'mob' extension cleanup.
• Un register storage controller VMs, and delete.  Confirm that the stCtl…. Folder was removed from local datastore
• Remove HX cluster in venter (make sure the esxi hosts, storage controller VMs, and HX cluster name is gone)
• Remove HX domain plugin  extension from venter MOB.  See cleanup doc for more details.
• Run HX customized ISO on all converged nodes to install fresh ESXi image
• Disassociate the service profiles from the 3 nodes.
• Delete UCSM HX org
• Re-run HX installer VM
• Make sure during install to check 'clean up ' disk partitions.

Thanks,

Kirk...

[IC2317]

Hi Kirk,

Yes, I am going to follow steps as mentioned in the link ( https://community.cisco.com/t5/unified-computing-system/hyperflex-customer-cleanup-guides-for-fi-and-edge/ta-p/3896668?attachment-id=174876 ). However, I am not sure if erasing UCSM configuration is necessary if I want to reuse the same UCSM. Its ok to delete suborganizations etc as mentioned in  step 7, but not sure if we can re-install without erasing UCSM.

[Kirk J]

All your pools, templates, policies,  are contained in the org,, so you don't need to do erase config of the FIs.

Just deleting the HX related org, will clear all those up.

Kirk...

[IC2317]

Hi Kirk,

Thanks. I have tried re-installing using mentioned steps, however Deployment via Intersight has failed at 50% with below error.

failed in Task: 'Check OVA checksum verification result' with Error: 'Unknown error occurred'

[Kirk J]

Usually that is issue with allowing the ESXi IPs for the servers access to 443 for svc.ucs-connect.com

Usually firewall issue.

Kirk...

[IC2317]

Hi Kirk,

Thanks. Now the new below error, not sure why these new errors are coming even after I am doing re-installation and previous installation succeeded. Can you please suggest on this one also.

failed in Task: 'Check Network for Storage Controller VM Configuration Result' with Error: 'Configure networking script generated an exception: Error while connecting to ESXi host. Please check the connection and retry'

[Kirk J]

Maybe try restarting the esxi host, and do retry.

Might need to open a TAC case.

Kirk...

[IC2317]

Hi Kirk,

Many thanks once again for your help. TAC recommended to do the cleanup again and run the installation via Intersight without configuring vCenter. Luckily after that, installation completed successfully and vCenter registration had to be done manually by running below command on one of the controller VM.

stcli cluster reregister --vcenter-cluster --vcenter-datacenter --vcenter-password --vcenter-url --vcenter-user

There are other customers also facing this issue and TAC confirmed that  installation succeeds without vCenter configuration and that was the case with me also. This adds manual step of adding vCenter later but  I think this is case with customers doing re-installation as first installation succeeded without any vCenter issue. 

[Kirk J]

I have had some cases where the errors call out ESXi connectivity, but in reality it is the vcenter reachablity.

Glad you got it squared away.

Kirk...