My ESX 3.0 Upgrade Part 1

My VI3 license keys have started filtering in. Seems every couple of days that I get some new keys from my support agreement, so it should not be too long and I will have them all. Today I upgraded my VirtualCenter database and application to 2.0. This went really smoothly after reading the installation/upgrade document from VMWare and the RTFM upgrade document. It seriously just was a matter of making sure everything was backed up and that the ODBC was specified correctly. I put in the few license keys I have and now we have a production VC2 server running.

The next thing on the list is to publish the VC client via metaframe and set up some delegation permissions for business owners so they can manage the applications and servers that they need to. Once this is done I will be moving full steam ahead with the upgrade. The thing with VI3 is that planning is the biggest part of the upgrade and it has to be right.

We are going with a staged migration approach from the current setup to the new setup. We have the hardware and the luxury of doing it this way. The only issues that I face at the moment is not having all my license keys to get some fresh hosts running. Once the license keys are in I can build fresh hosts for the migration and format some fresh luns. We will start with dev and test as we can do this during production hours and if all goes well we should have a documented process that we can provide to the change management board for scheduled outages on production servers.

We have also decided against the upgrade options for a number of technical reasons. One of them is our small root partition and the other reasons are outlined as possible gotcha's in the RTFM upgrade document on Mike Lavericks site http://www.rtfm-ed.co.uk.

At this stage the rough and basic guide I have is this. Of course there is much more to it than this and if you take a look at the documentation mentioned above you will see how much actually is involved.

1. Build new ESX 3.0 host and assign fresh luns for VMFS3 file system.
2. Backup exisiting VM's on ESX 2.x host.
3. Migrate VM's to ESX 3 host
4. Rebuild old host as ESX 3
5. Add rebuilt host and new host to new cluster in datacenter.
6. Rinse and repeat for other servers.

This is the basic process that we are following except on a larger scale with many more VM's and Servers. It seems simple in theory and once I have done it with the dev environment I will elaborate further with detailed information. The most work to me seems to be the creation of our new SOE for ESX 3.0 server, which will also be published as we get around to it.

Thats it for now, short and sweet. Oh and I have had some issues again today with the timekeeping on my esx 2.x hosts. We have decided the best workaround is to schedule an hourly cron job to keep the hosts in sync with the global ntp source. The time document on the site will be appended accordingly to reflect this solution.