If I'm planning on upgrading a server application, what is a good way of getting network access, while preventing duplicate network names or other side effects?
The current process I was going to try consists of the following:
- Shut down the original VM (call it
SRV) - Clone the VM (call it
TMP) - Restart
SRV(so service is resumed) - Reconfigure
TMPso the network adapter is isolated (disconnected, or on a test network) - Power on
TMP - Connect to
TMP(from the vSphere console for example, if it has no network) - Upgrade the software on
TMP - Shut down
TMP - Reconfigure
TMPso the network adapter is live again - Shut down
SRV - Power up
TMP - Rename VM
SRVtoOLD - Rename VM
TMPtoSRV
This assumes that there is no data being gathered by
SRVthat is needed during the switch-over period. If there is (log data for example, then this is probably not the right way to do things
If I want the server to be on the network after step 5, is my only option
- run
sysprepon it and give it a new SID - rename it
SERVER-TMP - change the IP to an unused address
Then when ready to go live:
- remove old server from domain
- rename
SERVER-TMPtoSERVER - change the IP to the production IP