Hi all, I have an upgrade to do and want to know if this plan will work and any other potential risks or issues I may encounter. As ever thanks in advance.
ESX/ VC implementation and upgrade.
VC Upgrade- Current VC version 2.01 Build 32042--Proposed VC version 2.5 Update 1
ESX Upgrade- Current setup 4 nodes ESX 3.01--Proposed setup 6 Nodes ESX 3.5
Proposed upgrade solution.
VC 2.5 to be installed on the current VC server located in the datacenter. For upgrades to ESX 3.0.1 it is a prerequisite to upgrade VC first. Multiple instances of VC can run on the same box.
ESX nodes 5 and 6 to be physically racked and stacked and located in chassis 3 as this has the highest levels of firmware and also provides redundancy across both chassis. Temporary ESX blade profiles to be created however the current setup with the ESX farm all located in one chassis does create a single point of failure. Therefore propose half of the nodes housed permenantly in chassis 3.
Installation of ESX 3.5 software . At this point there is still no requirement for licensing the ESX product so any delays with this will again have no impact. Addition and configuration of storage/networking adapter's vswitches etc as per installation and configuration guidelines.
Test connectivity and functionality of the ESX3.5- VC2.5 setup with no potential for firmware revisions to be a problem. We can run with this setup for a while with no impact wrt licences.
We currently run an n+1 configuration with many VMs powered off we have the required resources and capability to run all required VMs from 2 nodes within the existing cluster if necessary. At this point put one node in maintenance mode and remove the VMs.
Reboot the empty ESX node after applying the firmware upgrade. Readd this node to the existing 4 node cluster. Remove from maintenance mode and migrate test VMS to ensure no problems with HA/DRS etc. Once satisfied repeat this process to allow all remaining nodes to have current firmware revision.
Remove any VMs from selected node and remove from cluster thus freeing up one licence.
Bring one of the ESX 3.5 nodes online and add this to the existing cluster within the fully patched chassis (chassis1). This should be allowed as a licence has been freed by the removal of one of the existing ESX 3.0.1 nodes. Amend host files on all ESX nodes etc. Test HA/DRS by migrating newly created VMs for all current specs W2K3 - ST/ENT x86/x64 etc.
If this proves unsuccessful as there are some known issues with managing ESX 3.5 from VC 2.5 a second 2 node ESX cluster can be created which can be managed from the new VC2.5 server instance.
Each previous ESX 3.01 node will be in turn taken down. Upgrade to ESX 3.5. Add to new or existing cluster - dependency point 9. Test HA/DRS and start to migrate test VMs to ensure full functionality. Then start to migrate live production servers.
