Description of problem: While upgrading cluster from v4.5.16 to v4.6.8, the machine-api-controller pods stuck in CrashLoopBackoff state with machineset-controller container observing panic: ~~~ $ omg get po -n openshift-machine-api NAME READY STATUS RESTARTS AGE cluster-autoscaler-operator-7d798866b6-2qvwh 2/2 Running 0 1d machine-api-controllers-6db879646d-g8lmv 6/7 Running 302 22h machine-api-controllers-7c5d86bdf8-6rwtk 6/7 Running 272 20h machine-api-operator-588bbcbcd6-m6vmj 2/2 Running 0 5d ~~~ $ omg logs machine-api-controllers-7c5d86bdf8-6rwtk -c machineset-controller 2020-12-29T06:48:35.523967946Z I1229 06:48:35.523856 1 request.go:581] Throttling request took 4.054290308s, request: GET:https://172.30.0.1:443/apis/security.istio.io/v1beta1?timeout=32s 2020-12-29T06:48:35.644538709Z panic: runtime error: invalid memory address or nil pointer dereference 2020-12-29T06:48:35.644538709Z [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x141b406] 2020-12-29T06:48:35.644538709Z 2020-12-29T06:48:35.644538709Z goroutine 1 [running]: 2020-12-29T06:48:35.644538709Z github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1.getMachineDefaulterOperation2020-12-29T06:48:35.644617819Z (0x0, 0x0) 2020-12-29T06:48:35.644617819Z /go/src/github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1/machine_webhook.go:216 +0x26 2020-12-29T06:48:35.644617819Z github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1.createMachineDefaulter(...) 2020-12-29T06:48:35.644617819Z /go/src/github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1/machine_webhook.go:210 2020-12-29T06:48:35.644617819Z github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1.NewMachineDefaulter(0xc00051c900, 0xc0002b00e0, 0x17f39d8) 2020-12-29T06:48:35.644617819Z /go/src/github.com/openshift/machine-api-operator/pkg/apis/machine/v1beta1/machine_webhook.go:203 +0x65 2020-12-29T06:48:35.644637951Z main.main() 2020-12-29T06:48:35.644637951Z /go/src/github.com/openshift/machine-api-operator/cmd/machineset/main.go:2020-12-29T06:48:35.644653854Z 123 +0x656 ~~~ Actual results: - Upgrade stuck due to pods not being in Ready state. Expected results: - The upgrade should complete as expected.
We were relying on the following comment [1]: > // This value will be synced with to the `status.platform` and `status.platformStatus.type`. > // Currently this value cannot be changed once set. Having reviewed another BZ [2] assigned to a different team, their proposed fix will resolve the issue we are seeing here. Marking this as a duplicate as there is nothing for us to do. If the issue isn't resolved when the other BZ is merged, please reopen this. [1]: ttps://github.com/openshift/api/blob/78a19e96f9ebe77b8e0e6019f3cd3b4ae53b0fda/config/v1/types_infrastructure.go#L198-L199 [2]: https://bugzilla.redhat.com/show_bug.cgi?id=1890038 *** This bug has been marked as a duplicate of bug 1890038 ***