0.9.2 panic instead of cycle error

[gerr1t]

I just upgraded my terraform plans to 0.9.2 from 0.8.2. In 0.8.2 I encountered a cycle error when running a destroy, however, in 0.9.2 the cycle error is replaced with a panic.

Terraform Version

Terraform v0.9.2

Panic Output

Crash.log: https://gist.github.com/gerr1t/eea1fdc318f711dff17f3f98cd2edcfd

Releasing state lock. This may take a few moments...
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0x48 pc=0x159f196]

goroutine 1957 [running]:
github.com/hashicorp/terraform/terraform.(*State).sort(0x0)
	/opt/gopath/src/github.com/hashicorp/terraform/terraform/state.go:745 +0x26
github.com/hashicorp/terraform/terraform.WriteState(0x0, 0x8050700, 0xc4201052d0, 0x1038a15, 0xc420cbbd10)
	/opt/gopath/src/github.com/hashicorp/terraform/terraform/state.go:2000 +0x40
github.com/hashicorp/terraform/state/remote.(*State).PersistState(0xc420750060, 0x0, 0x0)
	/opt/gopath/src/github.com/hashicorp/terraform/state/remote/state.go:58 +0x79
github.com/hashicorp/terraform/backend/local.(*Local).opApply(0xc4201e48c0, 0x8075f80, 0xc42076a7c0, 0xc4201fc790, 0xc420831590)
	/opt/gopath/src/github.com/hashicorp/terraform/backend/local/backend_apply.go:138 +0x5d9
github.com/hashicorp/terraform/backend/local.(*Local).(github.com/hashicorp/terraform/backend/local.opApply)-fm(0x8075f80, 0xc42076a7c0, 0xc4201fc790, 0xc420831590)
	/opt/gopath/src/github.com/hashicorp/terraform/backend/local/backend.go:226 +0x52
github.com/hashicorp/terraform/backend/local.(*Local).Operation.func1(0xc4201e48c0, 0xc42064cb00, 0xc42064caf0, 0x8075f80, 0xc42076a7c0, 0xc4201fc790, 0xc420831590)
	/opt/gopath/src/github.com/hashicorp/terraform/backend/local/backend.go:246 +0x9a
created by github.com/hashicorp/terraform/backend/local.(*Local).Operation
	/opt/gopath/src/github.com/hashicorp/terraform/backend/local/backend.go:247 +0x192

Expected Behavior

Either the cycle error to be gone, or to present me with a cycle error.

Actual Behavior

A panic.

[jbardin]

Looks like the apply operation is returning a nil state during destroy, which panics in WriteState.

[jbardin]

@gerr1t,

I have a PR pending that will hopefully fix the crash you're seeing, but I haven't been able to reproduce your specific case yet. I have a feeling that you'll still have an error, probably the same cycle error, after this fix. If that's the case, it would be great if you could open a new issue with any info on how to reproduce the graph cycle on destroy. Thanks!

[gerr1t]

@jbardin Thanks! My expectation isn't that the cycle error is fixed, but that the exception was caught without causing a panic, seems like you fixed that, perfect!

[ghost]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues.

If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.