Server Upgrade : snapshot of PRD VM vs installing original software to new VM

[PeterPerera]

Hi 

We are preparing for an upgrade of server which is of version 2021.3.x.

Some suggests that better to go for a brand new VM to match existing PRD server. Install the server software (same version we have in PRD server) and take a copy of mongodb from current PRD and replace in brand new VM.

They say it is risky to try to get a VM snapshot directly from PRD or DEV so that we can run upgrades and test.

Is the above approach make sense?

This replacement of mogodb is that easy and will that work?

replacing mongodb will give same settings in PRD into brand new VM?

Please help.

[KOBoyle]

We're preparing for an upgrade from 2021.2 to 2023.2. We are planning to do an in-place upgrade. Our rollback plan is to stop the Prod server and clone the VM (retaining the machine name and IPaddress) prior to the upgrade. If there are any issues, we'll replace the Prod instance with the clone, and reschedule the upgrade for a later date.

We plan to do the following rollback test in advance:

1. Shutdown the Prod VM instance

2. Clone the Prod VM instance

3. Start the clone and confirm the Server and Gallery come up as expected.

4. Shutdown the clone.

5. Re-start the Prod VM

Assuming this test is successful, then we'll proceed with the in-place upgrade.

[fmvizcaino]

Hi @PeterPerera ,

This process is called Host Recovery. https://help.alteryx.com/20214/en/server/install/server-host-recovery-guide.html#server-host-recover...

Even though it looks like a complex process, if you follow the steps, you will have no problems.

I have posted numerous times in the server discussion board about this process and a list of things to be aware of when upgrading your server, such as doing the pre-upgrade checks. Take a look and let me know if you have any questions.

Best,

Fernando Vizcaino

[PeterPerera]

Thanks KOBoyle & fmvizcaino. 

New VM setup going to be done only to test the upgrade process from 2021.3.x to 2023.1. So a selected workflows currently in PRD will be tested in New VM, Scheduling  and all other functionalities also will be tested until we are satisfied with the upgrade process. One this upgrade process is successful we will upgrade the PRD server and test for weeks. The New VM used for upgrade testing will be decommissioned then. Is this sound correct.

[PeterPerera]

Basically trying to setup the "C" in the middle and upgrade as indicated in picture