I spend a lot of time patching my system. This is both public patches (to keep my reference systems ready for customer cases) and also a variety delivery vehicles (unreleased patches all the way to substituting loose files).

Increasingly, I found that I needed to keep a higher percentage of my terminal sesssions, either as exceprts or as full sessions in the wiki. After a major flurry of patching last month, I realized that I had not really been doing patching the way I wanted customers to do it. Here's what I do:

  1. Install patch w/ patchadd.
  2. Do any product-specific post-patch tasks.
  3. Poke the system to test the expected behavior (sometimes the change should be immediate, sometimes not, etc.)
  4. Restart the system and re-test.
What I realized, is that although this is "dogfood" (I use the stuff I give customers), I was not "wrapping" the patch activity with the information that I wanted. This in itself is interesting, because often customer-oriented techies assume that dogfood testing can be equated with the customer experience.

I should be checking the version with the product-specific method, and also using commpkg info. Before and after the patch session.

So, I'm starting to try to remember to do that now. Any comments on this? Let me know.

Read More about [Thoughts about versioning and my (personal) patch documentation...