core: Ensure EvalReadDataApply is called on expanded destroy nodes

[vancluever]

Ref: #6913

As per the discussion with @apparentlymart, I did a bit more digging in the code and the debug logs, and found the issue is actually with the ResourceCountTransformer. Log snip below:

2016/05/28 21:06:42 [TRACE] [walkDestroy] Exiting eval tree: provider.aws
2016/05/28 21:06:42 [DEBUG] vertex data.aws_ami_description.nat_ami (destroy), got dep: provider.aws
2016/05/28 21:06:42 [DEBUG] vertex root.data.aws_ami_description.nat_ami (destroy): walking
2016/05/28 21:06:42 [DEBUG] vertex root.data.aws_ami_description.nat_ami (destroy): evaluating
2016/05/28 21:06:42 [TRACE] [walkDestroy] Entering eval tree: data.aws_ami_description.nat_ami (destroy)
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalSequence
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalInterpolate
2016/05/28 21:06:42 [DEBUG] root: eval: terraform.EvalNoop
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalCountFixZeroOneBoundary
2016/05/28 21:06:42 [TRACE] [walkDestroy] Exiting eval tree: data.aws_ami_description.nat_ami (destroy)
2016/05/28 21:06:42 [DEBUG] vertex root.data.aws_ami_description.nat_ami (destroy): expanding/walking dynamic subgraph
2016/05/28 21:06:42 [TRACE] Graph after step *terraform.ResourceCountTransformer:

data.aws_ami_description.nat_ami (destroy) - *terraform.graphNodeExpandedResourceDestroy
2016/05/28 21:06:42 [TRACE] Graph after step *terraform.OrphanTransformer:

data.aws_ami_description.nat_ami (destroy) - *terraform.graphNodeExpandedResourceDestroy
2016/05/28 21:06:42 [TRACE] Graph after step *terraform.DeposedTransformer:

data.aws_ami_description.nat_ami (destroy) - *terraform.graphNodeExpandedResourceDestroy
2016/05/28 21:06:42 [TRACE] Graph after step *terraform.TargetsTransformer:

data.aws_ami_description.nat_ami (destroy) - *terraform.graphNodeExpandedResourceDestroy
2016/05/28 21:06:42 [TRACE] Graph after step *terraform.RootTransformer:

data.aws_ami_description.nat_ami (destroy) - *terraform.graphNodeExpandedResourceDestroy
2016/05/28 21:06:42 [DEBUG] vertex root.data.aws_ami_description.nat_ami (destroy): walking
2016/05/28 21:06:42 [DEBUG] vertex root.data.aws_ami_description.nat_ami (destroy): evaluating
2016/05/28 21:06:42 [TRACE] [walkDestroy] Entering eval tree: data.aws_ami_description.nat_ami (destroy)
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalOpFilter
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalSequence
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalReadDiff
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalFilterDiff
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalIf
2016/05/28 21:06:42 [DEBUG] root: eval: terraform.EvalNoop
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalGetProvider
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalReadState
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalRequireState
2016/05/28 21:06:42 [DEBUG] root: eval: *terraform.EvalApply
2016/05/28 21:06:42 [DEBUG] apply: data.aws_ami_description.nat_ami: executing Apply
panic: bad, not supposed to be here!

The node gets transformed into a graphNodeExpandedResourceDestroy which was not handling for data sources, putting all nodes through EvalApply regardless of what they were.

The fix was simple - handle the apply case properly with an EvalIf. Not too sure if an entirely new node type is necessary for this.

[jen20]

👍 This is a better fix than preventing fall through as I was trying to do. Thanks @vancluever!

[vancluever]

No problem and thank you @jen20!

[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.