Deployment on multiple test stations. Forward and backward integration of changes, keeping up to date.

[Paulius]

Hi,

I need some help and/or insights on how to manage a production test floor. Anything helpful will be appreciated!

Situation is this (numbers are just for illustration):

1. 10 identical test stations with mass interconnect receiver to have good reusability of standard T&M equipment.
2. 15 unique test adapters with their corresponding test applications.
3. Some test adapters have more than one duplicate to parallelize test process on multiple stations when needed.
4. any test can be deployed on any station or multiple stations.
5. some stations are stored as a backup.
6. All stations have PXI chassis with both NI and 3rd party devices, and other devices that are controlled through VISA.
7. We develop tests in such a way that root folder could be simply copied to identically configured station without using Deployment tool.
8. We have a centralized versioning system where latest software releases are kept.
9. Test are developed using TestStand, LabView and multiple other SW tools.

And then there are these issues:

1. Test application is updated with new limits (external limit file/db).
2. Test sequence is changed with new logic.
3. 1 and 2 can happen either off manufacturing on a development system or as an emergency fix on a station that is currently testing.
4. in 1-3 cases changes have to reach all affected stations at some point.
5. It must be ensured that production is not using an old version of test application (e.g. a station from storage is retrieved that has an old software on it).
6. It is not practical to manually go through all 10 (make it 50 or 100) stations whenever change occurs, because some stations are just a redundant hardware.

Are there any existing NI or 3rd party tools or solutions to manage this? Basically what I need is a way of controlling changes and delivering them to multiple test station. Knowing what station has what configuration is a must. Everything can be done manually (checking, deploying, book keeping, etc.) using spreadsheets, but manual work is error prone.

Is there an industry standard good practice or practices for this?

Thanks!

[~jiggawax~ · Accepted solution]

Have you seen this document: https://www.ni.com/en/support/documentation/supplemental/08/teststand-system-deployment-best-practic...

It is an interesting way to tackle the problem.  However, it is not an off the shelf solution like you are looking for.  You would need to implement some of the parts to make it work well.

I own a company that specializes in TestStand frameworks and architectures.  This issue is one we've been discussing lately.  I'm curious what you think some of the requirements a tool like this would need to have?

Our idea is that a database, or look up table, would link the part number to a set of "packages".  Then a thin client/GUI would be installed on the deployment machine.  It would then do a checksum against a shared drive repository of "packages" and what is local.  It would then download any deltas and then execute.  This way the user can dust off a test computer and it will update itself through the thin client.  You no longer have to manage the machines.

Regards,

[Paulius]

Hi jigg,

Thank you for your reply. I was thinking about exact same approach as you describe.

We already use a versioning system for our projects and do something similar to what is described in the document you attached, but manually.

I can also imagine other scenarios where having an equipment and project database like this would help with both typical and emergency management/support activities.

What I don't look forward to is starting our own in-house solution for this from scratch.