In 1.11 (1.32.0) we backported https://github.com/openshift-knative/serving/pull/515 which was first added in 1.12 upstream.
That PR created an issue that was later fixed in 1.12 and 1.13 (see for example https://github.com/knative/serving/pull/14795 for the fix).

Unfortunately as stated in https://github.com/knative/serving/pull/14795#issuecomment-1917540812 you need to patch every version that has the faulty PR because upgrade will fail otherwise.

While preparing 1.13 for 1.33 we noticed that we get failures when upgrading from 1.32 to 1.33 (see also https://github.com/openshift-knative/serverless-operator/pull/2578#pullrequestreview-1961424398).

The way to solve this is by patching 1.11 (1.32.0) or reverting the issue, creating a Serverless patch release, so that when the user upgrades from 1.32.1 -> 1.32.x or 1.32.1 to 1.33.0, future upgrades will work. We already have a PR that patches 1.11 with the fix, linked in the issue.

Unfortunately for upgrades from 1.31.x to 1.32.0 users will face traffic being dropped.
The latter means we need to notify users in the 1.32.1 release notes and about the existing bug when they upgrade from 1.31.x to 1.32.0.

One way to reproduce the bug, is to create a faulty webhook and dont allow pods to be created before the upgrade.