When rolling update is triggered by reconciliation trying to roll pods which cannot be started for example with following setting:

.withNewResources()
.withNewRequests()
.withMilliCpu("100000m")
.endRequests()
.endResources()

First rolling update is done correctly - timeouted after XX seconds with log from CO about it, but after next reconciliation strange debugs logs appears and CO doesn't show rolling-update timeout error again.