keda-operator crashs with SIGSEGV for kafka scaler

[flecno]

Expected Behavior

If the offset lag cannot be polled return 0 and do not crash with SIGSEGV.

Actual Behavior

keda-operator crashs with SIGSEGV:

panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x13c43ad]
goroutine 1544 [running]:
github.com/kedacore/keda/pkg/scalers.(*kafkaScaler).getLagForPartition(0xc0008f03c0, 0xc000000017, 0xc000978de0, 0x32)
	keda/pkg/scalers/kafka_scaler.go:281 +0x5d
github.com/kedacore/keda/pkg/scalers.(*kafkaScaler).IsActive(0xc0008f03c0, 0x21c2060, 0xc0017d0200, 0x1, 0x1, 0xc0016d2900)
	keda/pkg/scalers/kafka_scaler.go:168 +0xe0
github.com/kedacore/keda/pkg/handler.(*ScaleHandler).handleScaleDeployment(0xc0008aa630, 0x21c2060, 0xc0017d0200, 0xc000b96340)
	keda/pkg/handler/scale_loop.go:119 +0x11a
github.com/kedacore/keda/pkg/handler.(*ScaleHandler).handleScale(0xc0008aa630, 0x21c2060, 0xc0017d0200, 0xc000b96340)
	keda/pkg/handler/scale_loop.go:45 +0x6b
github.com/kedacore/keda/pkg/handler.(*ScaleHandler).HandleScaleLoop(0xc0008aa630, 0x21c2060, 0xc0017d0200, 0xc000b96340)
	keda/pkg/handler/scale_loop.go:14 +0x1d2
created by github.com/kedacore/keda/pkg/controller/scaledobject.(*ReconcileScaledObject).startScaleLoop
	keda/pkg/controller/scaledobject/scaledobject_controller.go:333 +0x35f

Steps to Reproduce the Problem

This problem occurs only when a kafka topic partition is not clean and ready to use. I have seen that when a broker as partition leader restarts.

Specifications

• KEDA Version: 1.2.0
• Platform & Version: GKE COS
• Kubernetes Version: v1.14.10-gke.17/v1.15.9-gke.8
• Scaler(s): kafka