Operator updates less aggressive restart of instance pods

Subscriptions offer the possibility to choose between automatic and manual update approval. Choosing automatic results in direct updates of operators as soon as the subscription is updated. In our case, automatic update approval results in updating operators. Which, in turn, start restarting pods in our Match 360 instances. This may lead to (temporary) reduced service or unavailability.

The initial work-around is to set the approval to manual. In this case the administrator can select the actual moment of Subscription approval. However, we've seen it several times that the Subscription pages showed no pending approvals but nothing got updated as well. It turned out that underlying InstallPlans were still waiting for approval. But without especially looking for it, no operator update happens and subsequently no instance update became available. Manually approving the InstallPlans (order???) didn't always result in updateable instance.

The real workaround: keep it automatic.

However, updating the subscriptions and associated operators automatically also results in temporary reduced service or unavailability. In some situations (production or acceptance environments) this isn't an acceptable situation. We want to be able to prepare updates and execute them at our convenience. Furthermore, an updating operator shouldn't initiate a restart of its controlled services when no restart is needed (the service itself isn't eligible for an update yet, so why restarting pods, it is running fine).