Closed sudermanjr closed 5 years ago
Would we want to do a migration? It seems like it would require a very conscious choice from the operator to make that change and apply it to the kops cluster. Not sure if that falls under the purview of migrations, but I've also not ever written one before....
The area between "make a migration" and "enforce standards externally" is a grey one and we haven't really nailed it down. I think that any change in pentagon
standards deserves a migration and things that are outside of pentagon
are for external enforcement. There is certainly lots of room for discussion on this though and for halfways too. For example, the migration can make the changes and the user can reject them by not merging the migration
branch then the external audit tooling can nag them about it forever more.
I don't think that discussion should be a blocker for this issue but I think it will be a good thing to settle before a new release is cut so we can add a migration if the discussion falls that way.
That seems fine to me. I can stall this PR if you want me to add a migration, or we can merge this and I can add a migration. I also don't want to block any releases that might happen before I have time to write a migration
Merge this and create a new issue for a migration please. We can hold the migration discussion on that issue.
EJ Etherington Chief Technical Officer ReactiveOps.com ej@reactiveops.com
On Nov 19, 2018, at 8:22 AM, Andrew Suderman notifications@github.com wrote:
That seems fine to me. I can stall this PR if you want me to add a migration, or we can merge this and I can add a migration. I also don't want to block any releases that might happen before I have time to write a migration
— You are receiving this because your review was requested. Reply to this email directly, view it on GitHub https://github.com/reactiveops/pentagon/pull/151#issuecomment-439952277, or mute the thread https://github.com/notifications/unsubscribe-auth/AAp5zBq8SDJzIe0iS7-xlusd4HPpMFJzks5uwtrFgaJpZM4YVhF5.
This has been our default for a while. Sets up one instance group per AZ for a kops cluster