Upgrading CallManager from 3.2(2c) to 3.3(3)

Unanswered Question

Our customer wants to upgrade their CallManager cluster from 3.2(2c) to 3.3(3). One publisher and one subscriber on MCS 7835s.


I have two questions regarding the upgrading guide documentation:


(1) Why do we have to have two hard drives on the same server when upgrading? For the sake of safety, I want to remove one hard drive to keep the current version of callManager(3.2(2c)), and use only one hard drive for upgrading. Can I do this? Because on lab environment, it is okay. But from the documentation, it is required that the server should have two hard drives when upgrading.


(2) After upgrading the publisher callManager, can I just fresh new install the subscriber instead of according to the guide documentation in case I remember to update the hosts file about the publisher callManager and the subscriber callManager's IP adresses and the hostnames?


Thanks,


Joe

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
adignan Tue, 12/02/2003 - 19:12
User Badges:
  • Blue, 1500 points or more

1. Why not pull a drive from the subscriber. Pull a drive from the publisher, insert the subscriber drive into the publisher. Let it mirror. You will then have a working drive of the publisher if it makes you feel safer. I believe if you build it with only one drive in it wil change to RAID 0.


2. Yes, you can just do a fresh subscriber install. I have done this multiple times with no ill effects.


adignan - berbee

jasyoung Tue, 12/02/2003 - 19:13
User Badges:
  • Gold, 750 points or more

1) The reason is that the MCS imaging process sets up the RAID array controller configuration for you. If you run through the build with only one drive, you end up with no RAID and no way to get it without rebuilding from the ground up. I have a process that gets me through the upgrade while being able to back out of the important bits:


- Pull a drive from the publisher

- Pull a drive from the subscriber

- Insert subscriber's drive into the publisher


The subscriber's drives and other info not stored in the database are somewhat disposable, so we can get away with this. We keep the publisher drive set aside for a while so we can back out the publisher upgrade.


- Wait for RAID drive rebuild to finish

- Upgrade the publisher


We have two drives in it at this point, so the reimage goes fine.


- Stop the subscriber or otherwise flip everybody's service over to the publisher

- Make sure everything is stable and otherwise happy


Don't proceed from this point until you're sure you don't need to back out from the publisher upgrade.


- Insert publisher's spare drive in subscriber

- Wait for RAID rebuild process to finish

- Upgrade subscriber

- Life goes on normally from here


You can't back out of a subscriber upgrade with this process, but that's not a serious concern IMHO.


2) I've been asked this question a few times. This is of course feasible, but is usually unnecessary. If you go through the upgrade process, you preserve a lot of things, far more than just IP addressing - you resurrect the server into its former role in your IPT system. If you reinstall, it will be a new server so far as your cluster is concerned, so you will have to rearrange its resources back into your CallManager groups, media resource group and media resource group lists, service parameters and activations, etc. If your configuration is simple, this might be acceptable, but really it's not necessary - the subscriber upgrade process is very easy and not as prone to failure as the publisher.

Thanks for your information.


But the problem is that the Publisher callManager has two 18.2G hard drives(old MCS7835), and the Subscriber has two 36.4G hard drives.


And the customer wants to pull one hard drive before the upgrading for the safety. I am thinking that we lend one spare 18.2G hard drive for the publisher upgrading, and one 36.4G hard drive for the subscriber upgrading to customer for a couple of days.


After finish the upgrading, and verified the callManager's functionality. Say for a few days later, can I pull the drives which we lend to customer, and insert the customer's original publisher hard drive(3.2.(2c)) to be mirrored as 3.3(3) on the slot where I put our hard drive into during the upgrading?


If it not works, I want to do like this:


Publisher:

Pull one drive from the current publisher

insert our drive into it

mirror it

pull the another drive from the current publisher to test only the mirrored drive works fine

If the mirrored hard drive works fine, pull this drive and insert the customer's original two hard drives into slots and begin to upgrading with the two customer's hard drives.


Thus we use our drive to backup the customer's current version of CallManager.


Subscriber:



Pull one drive from the current subscriber

insert our 36.4G drive into it

mirror it

pull the another drive from the current subscriber to test only the mirrored drive works fine

If the mirrored hard drive works fine, pull this drive and insert the customer's original two hard drives into slots and begin to upgrading with the two customer's hard drives.




Thanks a lot.


Joe

Actions

This Discussion