kubevirt / hyperconverged-cluster-operator / 28793681455

82%

Build Type

push

github

Committed by web-flow

Commit Message

Fine tuning the MDev's enabled field and FG (#4376)

* explicitly don't auto-convert deprecated FGs

We already make the pkg/featuregatedetails/feature-gates.json file to be
the source of truth for the feature gates, instead of the v1beta1
`HyperConvergedFeatureGates` field. But the fg-json-generator tool still
uses the old source. In addition, we no longer auto generating the json
file.

This commit modifies the fg-json-generator:
* rename to the tool to `fg-conversion-generator`, to reflect
  the changes
* remove all the json-generation related code from the tool
* add comments in the generated file, for better readability.
* sort the feature gates in the generated file (by phase,
  then by name), for better readability.
* explicitly not convert deprecated FGs. Add comments instead.

Signed-off-by: Nahshon Unna Tsameret <nunnatsa@redhat.com>

* add a method to check if a FG is explicitly set

Signed-off-by: Nahshon Unna Tsameret <nunnatsa@redhat.com>

* Change the conversion logic for the MDev.Enable

V1 => v1beta1:
```
Use the v1-only-fields mechanism to store the Enabled field and v1 FG
If enabled == nil => v1beta1 FG is nil
If enabled == true => v1beta1 FG is false
If enabled == false => v1beta1 FG is true
(v1 FG is not converted)
```

V1beta1 => v1
```
V1beta1 FG is not auto-converted => v1 FG does not exist (yet)
Restore Enabled and v1 FG using the v1-only-fields mechanism
If v1beta1 FG is not nil:
  override restored Enabled = not(v1beta1 FG)
  If restored v1 FG is not nil => v1 FG = v1beta1 FG
```

Signed-off-by: Nahshon Unna Tsameret <nunnatsa@redhat.com>

* Fine tune the sync of the MDev enabled field and the FG

Fix several issue, bugs and behaviors in the logic around syncing the
disableMDevConfiguration FG, anf the
mediatedDevicesConfiguration.enabled field.

For example:
* never remove the enabled field if already set.
* when dropping the FG, if it's the only FG in the list, remove the list
(to restore the default value).
* differentiat... (continued)

Coverage Stats

126 of 134 new or added lines in 3 files covered. (94.03%)

8 existing lines in 2 files now uncovered.

11284 of 13845 relevant lines covered (81.5%)

2.19 hits per line