Migrate UC VM's from UCS C220 server to BE6k server

[Ciscollab_Amit]

Hello guys,

I need to migrate a customer's UC infrastructure from UCS C-220 to BE6K server.

what would be a good recommendation incase no downtime is needed during the migration and both the old and new systems should run parallely. Once the new BE6K is completely ready and tested, I would then like to just shut down the old UCS C-220 Box. The customer would also like to have the same IP addresses for the new system as it was on the old system. The ISO is available on the local UCS harddisk and not on the NAS.

But what I am planning is as below..

1. Connect the new BE6k in the DC and configure BIOS, CIMC and network config's for the new BE6k.

2. Install the PCD VM on the BE6K ESXi host and do the old UC component discovery.

3. What should be done in this step? Shall I install empty VM's with OVA's, that are already available in the BE6K datastore with ver 11.x? The old UC infrastrusture is on 9.1 platform.

Currently I am stuck at this point. The current environment has below applications:

CUCM-Pub ver 9.1

CUCM-Sub ver 9.1

CUCM IMP pub and Sub ver 9.1. Connected to the Microsoft Lync.

VG migration from 29xx to 4000 ISR series routers.

Thanks for the feedback.

Regards,

Amit

[Jaime Valencia]

https://youtu.be/2C_gy_XmBmE

[Ciscollab_Amit]

Hi Jaime,

As always that was a quick one. :) So incase I need to test that the migration was really successful, I just need  to prepare the test plan and carry it out during the non-business hours. And in case of any catastroph, I just need to power off the new BE6k and I can still power on the old UCS box right or is that not an option? Also, I believe I do not need to compare the the ESXi versions between the two server's (UCS-C220 and BE6K)?

Is there also any tool available to check that the current UCS-C220 is system ready to be migrated using PCD. Because I have heard from lot of consultants that they have faced challenges using PCD for Migration.

Thanks for your feedback.

Regards,

Amit

[Jaime Valencia]

Yes, if you configure your separate network and use NAT you can use the same IPs and then just connect / disconnect.

There are no tools to find out if PCD will work or not, plenty of times the problem is not even in PCD, but in the source cluster or the settings for the new cluster install.

[Ciscollab_Amit]

As you said "plenty of times the problem is not even in PCD, but in the source cluster or the settings for the new cluster instal."

Could you give me some more pointers what could be the issues in the source cluster that could impact PCT migration or what could be the settings in the new cluster install that could impact the migration.

[Jaime Valencia]

Source cluster, problems in the DB that would also prevent an upgrade. Way too many options, those you can only track down via the upgrade logs and the DB logs to find out the specific entry in the DB that's causing the problem.

Install of new cluster, pretty much any problem you would have on a regular install, DNS, NTP, etc.

[Ciscollab_Amit]

Good to know.. Atleast I can verify the existing Databases before starting the PCD migration. Incase there is no ip address or host name change on the new Install cluster, I believe there should not be any network problems. If the ESXi versions, COP files and OVA files on the old and new install is not an issue, then hopefully, the migration should be smooth.

Incase I do not change the IP and if there is an issue with the Migration after switching, can I just turn off the BE6k and do a Disaster Recovery on the old UCS C220 system again? Should that help? Or even the disaster recovery is not needed and just turning off be6k and turning on UCS-C220 will work?

Regards,

Amit

[Ciscollab_Amit]

Hi Jaime,

So I just changed the topic. This is because I went onsite and then I figured out that it is not a UCS but a MCS server .:) but the second thing now is customer has different OS admin passwords for CUCM and IMP. When I discover the old cluster, PCD discover's CUCM and prompts credential error for IMP. So I am now trying to recover the OS admin password if possible. If not then I will be creating 2 tasks 1 for CUCM and 1 for IMP cluster discovery. As of now I cancelled the cluster discovery task for CUCM. But I believe PCD would have created the COP file and installed it on the CUCM. shall I delete this file from CUCM CLI, is that possible, or can I just recreate the discovery task and the COP file would be updated automatically with this new cluster discovery task.

Thanks for your feedback.

[Jaime Valencia]

You cannot remove a COP file once it has been installed, if you mean the COP file PCD installs for discovery, it will be used for any discovery procedure you might do.

[Ciscollab_Amit]

Yes, I meant the COP file that PCD installs after cluster discovery. So there is some more information that I receive after Cluster discovery.

CUCM(PUB and SUB)- They are currently on MCS hardware

IMP(PUB and SUB)- They are on VM

Did a Cluster Discovery and PCD was unable to discover IMP because IMP has different OS admin password than CUCM. In the same discovery task, I did an edit for the IMP (PUB and SUB) and changed the OS admin password to the one set for IMP and did the refresh node. This time it discovered the IMP as well. Now just to make sure the COP file has the correct data Synchronized across all the nodes, I did a refresh node on all the nodes in the Cluster. The discovery status is successful. Now when I try to create the Destination Migration Cluster, I am trying to assign the VM for CUCM Pub on ESXi-A, at the same time the PCD is assigning the CUCM-sub also to ESXi-A. But I want to assign CUCM-sub to ESXi-B. PCD is not allowing. PCD automatically assigns the CUCM's to the same ESXi host. What could be the issue here. When I do a reassign and assign the ESXi-B to subscriber, then the CUCM is also automatically assigned to the ESXi-B.

Next, I also have a question here as the Cluster discovery is a mix of MCS for CUCM and VM for IMP with different OS Admin Password. Will it lead to a successful migration still? or do I take the manual route for Migration. 

Currently I am referring to this document for Migration  https://www.cisco.com/c/en/us/td/docs/voice_ip_comm/cucm/pcdadmin/11_0_1/CUCM_BK_P00E3B55_00_pcd-admin-guide-115/CUCM_BK_P00E3B55_00_pcd-admin-guide-115_chapter_011.html#CUCM_RF_U171517A_00

I am doing PCD M1 migration. It unfotunately gives no way to migrate directly to the latest version of CUCM and IMP. I need to migrate to 11.0(1) first and then to version 12.x

Regards,

Amit

[Ciscollab_Amit]

I have an answer to my own question. This is strange that PCD cannot determine 2 VM's on different ESXi host with the same VM name's. I kept the default VM name that came alongwith the BE6K box which was Cisco Unified Communications Manager and Cisco Unified IM and Presence. I changed the name of the VM's and then I could choose what VM I need to install on which ESXi.

More to come... I was not able to select the source Cluster from the drop down list when I was creating the Destination Migration Cluster. But when I typed the first few letters manually,  it did recognize the cluster and pulled all the information for the source cluster. This also looks strange.

The question still remains the same, after carrying out these workarounds, will the PCD migration still be successful. 

Would be good to hear from you guys..

Regards,

Amit

[SalimMach]

Hello Jaime,

The video is unavailable at the moment,

can you please share another link if possible or the steps briefly?

thanks

[Jaime Valencia]

Video is not available anywhere else, I do not provide all the steps, just provide a brief explanation of the options in the video, with quick pros/cons:

A) DRS

B) PCD (if upgrading or moving to a higher release, even an SU above would work)

C) Cold migration (with vMotion or manually)

D) Shared storage migration

E) Copy/clone the VM

F) Move files directly between servers with SCP or SFTP via CLI

G) COBRAS if you have CUC