Adding Stickiness option for Target Groups

[drmudgett]

Adds the option to enable Sticky Sessions to the ALB, defaults to false (off).
Closes #267

[szuecs]

All golden files tests would have a CF stack change because of the new parameter. I am not sure if you can do this such that default is nil instead of "false", then you don't need to update tests.

[drmudgett]

All golden files tests would have a CF stack change because of the new parameter. I am not sure if you can do this such that default is nil instead of "false", then you don't need to update tests.

I've updated the tests and they're passing locally now, thanks.

[szuecs]

All golden files tests would have a CF stack change because of the new parameter. I am not sure if you can do this such that default is nil instead of "false", then you don't need to update tests.

I've updated the tests and they're passing locally now, thanks.

We have to check if we can allow this.
Best would be if "false" that there is no parameter change in the CF stack.

The problem that might happen is that cloud load balancers could be forced to be new created.
We had a similar problem in the past that lead to downtime of some production applications.

[drmudgett]

All golden files tests would have a CF stack change because of the new parameter. I am not sure if you can do this such that default is nil instead of "false", then you don't need to update tests.

I've updated the tests and they're passing locally now, thanks.

We have to check if we can allow this. Best would be if "false" that there is no parameter change in the CF stack.

The problem that might happen is that cloud load balancers could be forced to be new created. We had a similar problem in the past that lead to downtime of some production applications.

Ok, I think I've got it working the way you'd like now.

[szuecs]

I guess these fields should not be there if we have not sticky configuration.

[drmudgett]

I've removed the line that the linter was complaining about, but I still see this error locally:

Error: kubernetes/pods.go:103:9: S1009: should omit nil check; len() for []k8s.io/api/core/v1.ContainerStatus is defined as zero (gosimple)
	return p.Status.ContainerStatuses != nil &&

But it looks like that error was there before my commits (i locally tested against main).

[szuecs]

that's a new staticcheck version, so would be great if you just fix it :)
https://staticcheck.dev/docs/checks#S1009

[drmudgett]

no problem, this is fixed in the latest commit

[szuecs]

but the result params should not change and as far as I read the code.
This should not be there in the PR:

},
  {
    "parameterKey": "Stickiness",
    "parameterValue": "false"

Ah and please add a test that actually tests the feature with "true" and "false" set :)