I have to upgrade a virtualized CUCM cluster (PUB, SUB1, SUB2, TFTP) version 8.6.2 SU3 to version 10.5.
As per documentation this should be a direct refresh upgrade and no need of PCD to do it.
Based on the documentation the main steps of the upgrade are:
-changing the OS to RHEL 6 (64-bit) and the Network Adapter from "Flexible" to "VMXNET 3” for the VM (shut down the VM and perform the changes)
-run the 10.5 ugrade file in following order: PUB, TFTP, SUB1, SUB2
The existing 8.6.2 OVA, 7500 users is using 2 vCPU and 80GB for each vdisk (total of 160GB).
The 10.5 OVA, 7500 users requires 1 vCPU and 110GB for the vdisk.
Does anybody know if I have to run the ciscocm.free_common_space_v1.1.cop.sgn to increase the space to 110GB on each vdisk (even if the total size is 160GB on 2 vdisks which is more then 110GB required on 10.5 OVA)?
Cisco engineer said that this is not possible as well that I can not change the OS to RHEL 6 (64-bit) on the existing server/VM (ESXi 4.1 U3.
He recommends to create a new VM using the new OVA and use the DRS and restore process for performing the upgrade.
Any input/advise will be appreciated!
Solved! Go to Solution.
The success to any upgrade is planning. The customer type bank or not doesn't not really matter. What is crucial is your willingness to do due diligence to make this as smooth as possible..
Here are the things you need to do in the order described
1. Read your documentation. Very crucial. CUCM10.5 is a major change on all others.
You need to start by reviewing the upgrade guide, release notes, and new and changed documentation for all the steps and considerations you need to take into account for this upgrade.
NB: The upgrade guide for IM and P is merged with that of CUCM and here is the link for it..
Release notes found here:
2. Plan for your license migration. Usually this involves two steps
a. Use cisco LCU (license count utility tool) to generate license report on your existing cluster. This can be downloaded from cco.filename is LCUZIP-Ver9.1.2.zip
Ensure that you install any unsed DLU and license PAKS on your cluster before you run this..
b. Next email this file obtained from lcu (this is a csv file) tool cisco licensing team firstname.lastname@example.org. This will confirm that your service contracts is in order prior to requesting new licenses for 10.5
c. Login in Prime License Manager after successfully migrated from 8.6 to 10.5
Use the following procedure to generate a license request.
3. Document the steps you will use for the upgrades. You can send a draft here and I will be happy to review this with you.
Ensure that you write down step by step what you will do on the day. The upgrade guide should be used to write down what you will do on the day. This is a very long upgrade, hence you need to document in detail what you will be doing.
4. Contact the relevant team to get your ESW/UCSS contract numbers. You need this to login to PUT to download your media. I was doing an upgrade recently and realised that the customers ESW contract has expired..So this is crucial.
PCD is your best bet. Install PCD on your Esxi host. Just create a new VM for it. It makes things much easier for you..If you don't have spare capacity to create new VM then do it manually
5. Before you begin..Take a back up of your cluster.
6. VM machine sizing.
Please refer to cucm10.5 ova readme file, found on CCO. Depending on the OVA template you used for your cucm8.6, you may need to change the vm setings prior to upgrade. This is because the disk size and cpu requirements is different for cucm10.5 on some ova templates.
6a. You will need to change the vm adapter settings on your VM. This is documented in the cucm10.5 ova. Your best option is to use esxi powercli. You can download this from VMware website provided you have a valid service contract.
6b. You will need to change your guest operating system to RHEL6. This is also documented in the 10. 5 OVA
7. If your 8.6 cluster is running on VM at the moment, then just do a direct upgrade to 10.5.
NB: before you upgrade you will need to install the ff cop file on all the servers in the cluster
After this upgrade your cluster starting with cucm. The upgrade guide will tell you the order of the upgrade..
7. Once upgrade is completed, perform the post upgrade tasks.
8. Develop a detailed test plan. Ensure all critical component of your customer is covered in the test plan. Test features that are critical to them
9. Finally, ensure your CV is updated. In case you need to look for another job if everything breaks: (lol)
Answer to your questions:
"As per what I see, an easy option would be to create a new virtual machine and use a DRS? but, would it work if I create the DRS on CUCM 8.6 and use it on 10.5?"
---You may create a new VM or adjust VM settings as I already discussed. If you want to create a new VM, then you will create VM with cucm10.5 ova, install cucm 10.5 on it.
You will then upgrade your current 8.6 to 10.5, take a DRS back up and go and do a DRS restore on the new 10.5 you have installed on the new VM. You cant restore 8.6 to 10.5.
"What happens to presence server? I understand I have to upgrade it too but, will the configuration be transferred to CUCM 10.5 (I mean, services associated with the menus that originally were on presence 8.6)?"
---Yes IM and P needs to match the cucm version. Yes the configuration will be transferred. You don't need to do anything. Please read the upgrade guide carefully.
"If all of the apps are on 8.6, do we need to upgrade everything to 10.5 at the same window?"
---For CUCM and IM and P, yes you have to upgrade them at the same time. The versions must match. It is advisable that you have unity connection on same version too, but not critical. I believe 8.6 unity connection will work with cucm10.5 ( I haven't checked this, so confirm it yourself)
The upgrade is a refresh upgrade. So it does take a long time and there are several reboots. The publisher is the one that takes the longest amount of time. The subscriber is about half the time it takes for the Pub to upgrade. You should be fine with a weekend maintenance widow. Friday evening to Sunday.
"If upgrade fails, can I still go back to 8.6?" Yes as long as you do not use the "free common space cop file" This is discussed in the upgrade guide as well.
It looks daunting, but if you do your home work, you will be fine and we are here to help and assist you all the way
Please refer to this thread for similar discussions:
Last week I did an upgrade to CUCM 10.5 using PCD. I am not sure what you mean by no need for PCD to do it. PCD is a great tool and it can make your life a lot easier if you do your home work very well and by your home work I mean do your pre-upgrade tasks as documented by cisco.
One down side to PCD is that you don't know what is going on unlike the manual upgrade where you can see logs and can see if a failure happens why it failed etc.
My upgrade kept failing and I had to assume control (take over PCD) upgrade to see what the error was and the error was common space partition issue..(once I corrected this, my upgrade went smoothly)
Back to your questions..
First thing to note is this, the free common space cop file has a big downside. It wipes out your inactive partition. SO essentially you will have nothing to fall back to if anything goes wrong with your upgrade other than to do a DRS. So this should be a last resort. Here is a short note from the read me file..
What you should do, which is what I did is to use RTMT to free up space in your common partition and to also adjust your "loglevellowwatermark and highwatermarklevel" This is what cisco's upgrade guide recommends..
If I were you, I will use this opportunity to upgrade my esxi host as well and eploy new OVA and then do DRS as suggested by TAC. This way are not doing much fiddling and you can get your servers up and running with the best specs..
I must also add "refresh upgrade are service impacting" There will be lots of server reboots etc and it takes a long time. So plan accordingly
Thank you for your answer Ayodeji,
The problem I face is that I do not have enough resources on the UCS servers to deploy the PCD and to create additional VMs to use for upgrade.
Based on this the only option I see is to do a manual upgrade which is an option as per the documentation.
Did anybody run the ciscocm.free_common_space_v1.1.cop.sgn? Any impact on the system?
Just add to my earlier post, I can confirm that you can run RHEL 6 (64-bit) on the existing server/VM (ESXi 4.1 U3.) I am running 4.1.0 on one of my host vms and I am able to change the OS to RHEL 6(64bit)
I'm facing the same issue.
My customer wants to upgrade from CUCM/Presence/Unity Connection 8.6 to 10.5.
Honestly, this is the first time I'm going to do a major upgrade and I'm very worried about it, specially because the customer is a bank.
1. I've been looking for information but I'm very confused. If customer doesn't have Prime Collaboration, then, can't I upgrade manually?
2. As per what I see, an easy option would be to create a new virtual machine and use a DRS? but, would it work if I create the DRS on CUCM 8.6 and use it on 10.5?
3. What happens to presence server? I understand I have to upgrade it too but, will the configuration be transferred to CUCM 10.5 (I mean, services associated with the menus that originally were on presence 8.6)?
4. If all of the apps are on 8.6, do we need to upgrade everything to 10.5 at the same window? I mean, I was reading that a server can take up to 4 hours to complete the upgrade process, we have:
CUCM: 1 pub and 3 subs
Presence: 1 server
Unity Connection: primary and secondary server
CUCM SME: 1 pub and 1 sub.
We need to schedule a maintenance window but, if everything has to be done at once it will be a huge window!
5. If upgrade fails, can I still go back to 8.6?
Thanks a lot in advance!
if u are doing major upgrade , then u would have ordered for CUCM 10 through PUT and u would have received PCD media.
secondly, backups are platform independent.
Thanks to Ayodeji Okanlawon and Aman,
I still have some doubts please clarify , I have a Customer using 8.6.2 in Virtulized environment, the initial deployment of 8.6.2 was not done by using ova's, so if i do refresh upgrade, i think it will result into "ERROR-UNSUPPORTED" Partition Unaligned....
The Question is can i deploy new 10.x Ova's for CUCM-PUB,CUCM-SUB1,CUCM-SUB2 and CUP-SUB and use PCD to do migration from Old Cluster to New Cluster with same IP Address and Host names, is Customer eligible to have new Images of 10.x or he will get only upgrade images which can't be used to migrate?
If i am doing migration by deploying ISO images in Fresh_install directory of PCD, can i use the same upgrade images which i downloaded using PUT for upgrade.
The customer has all UCSS Contract and ESW Contract
Are u referring to Bootable images of CUCM 10.sgn.iso which get installed in fresh_install[SFTP datastore of PCD]? the same u would have downloaded from Cisco after placing order through PUT
Thanks Aman.......... Yes i am referring Bootable images of CUCM 10.sgn.iso
Q) If customer orders using PUT is he eligible to download fresh Install ISO images for CUCM, CUC and IM?
and how about license migration ?
All the information above are of tremendous value to me as i am going to do the upgrade of cucm 8.6 to 10.5 so my +5 to you all. I have two questions as I go through the upgrade guide.
1. Under the section "refresh upgrade of cucm nodes only..." it says switch version in pub after the upgrade and before starting upgrading subscribers. However, in another section called 'parallel upgrade' in the same document, it says upgrade all the nodes before switching versions. It has confused me. I want to upgrade only the CUCM nodes this time. Only application I have running is CUC that i will upgrade at another time.
2. I have 28 GB free space in common partition in pub and way more in the subs. My old OVA template is for 7500 users (2 vCPU,6 GB RAM, 2x80gb vDisk). Should I still need to increase the disk size from existing 80 gb? If so how?
Thank you, Khanal
Have you done your upgrade yet? Which order did you switch your servers? There seem to be conflicting information about the other of this for a refresh upgrade..
I have the same question with the existing 8.6.2 OVA, 7500 users is using 2 vCPU and 80GB for each vdisk (total of 160GB), can I power off the virtual machine and increase the storage and change the Os to RH 6 64x.
I have 2 UCS wit 1 Pub/ 1 Sub /1 Voicemal.
I will deploy a Ova10.5 and install the 10.5 iso on the machine. My question is if I can restore the back-up from Cucm8.6 to new machine 10.5 and power off the 8.6 and then change the ip and dns on the new machine.
As per DRS doc you cannot recover a different version backup.
You can install a new 8.6 sever, recover from DRS and then upgrade to 10.5
After upgrade you can change ip address and DNS .
Carlo (+5). How are you doing..
I'm doing great thanks my buddy. :)
Hope to see you in Milan next January
If this can help more our friend, I reproduced an upgrade from 8.6 to 10.5 in my lab and our suggestions are quite all you need.
1. Finally I have all the pack for upgrade. I have 10.5 OVA template and I upload on vmware and tray to install the cucm10.5 on this template but I receive the error from attachment.
I made a static reservation for this machine Ova 10.5 bot working, I place a cucm9.1 ISO and works fine.
I will upgrade this Saturday and I need to install this machine.
Any ideas why i receive this error?
The second question is about update 8.6 to 10.5
I tray to update my cucm8.6 in home Lab but after 10 hours still installing, maybe i have to do something extra.
I upload first the version3 then UCOS_UNRST_10.5
My question is what is the proper sequence in order to make update from those files:
Most probably because that iso is non bootable.
You can use a bootable image an extract the boot info to than upload it to that ISO.
I have 8.6 Unrestricted on the vmware workstation 10 and I use chrome/Firefox and internet explorer but the same problem. I have to upload to Pub and Sub the ciscom.refresh_upgrade_v1.5.cop.sgn or order to be able to update direct to cucm105 or only the ciscom.version3-key.cop.sgn or ciscom.free.common_spacev1.1.cop.sgn
Thanks for replay
Please go back to the answers I provided earlier in this thread. You should take time to read your upgrade guide documentation before you start doing this. Why are you installing the refresh upgrade COP file? Do you need it? Have you read the documentation on which versions of cucm you need this? Why are you still talking about the free common space partition cop file? Do you need it? I already told you this earlier and I don't know why you are still referring to it. Why did you install the UNREST cucm version? Who advised you to install this? Is this the version you are running in your production system?
You have so many threads opened on one subject. Please focus on one thread, so we don't have to answer in several places.
Read the upgrade guide for cucm10, read the answer I provided above and you will get a better picture of what you need and do not need.
Thanks for replay and sorry for bad question.
I have Unrestricted because my subcontractor installed unrestricted on the server
I manage to update my cucm8.6 to cucm10.5 but now i receive after update the error that I have unaligned partition and to introduce the recovery CD.
I am not able anymore to log in to GUI.
This is what i done:
I not use any template. I manual config the cucm8.6 like in production environment and then upload the cucm10.5 Unrestricted and after update i receive the error. (unaligned partition)
Or I don't understand the question?
Th for replay
Is this the way u did?
a. Took back through DRS on existing VM
b. performed migration and then, restored.
Can u share snapshot of error message?
Try and rebuild your db replication with the following command from the publisher..
utils dbreplication rebuild