`pulumi preview`/`pulumi up` are both giving inconsistent previews. I’ll run

pulumi preview

and it will say that I have 1 resource to delete from AWS, and then I’ll run it a few seconds later and it will report that 7 resources need to be deleted. Then I’ll run it again and there will be a completely different set of resources it wants to delete. This is all without me making any changes the the underlying resource definitions at all. Interestingly this is only happening with

aws:glue:CatalogTable

resources — my other resource types don’t seem to be impacted. My state is local in

./.pulumi

and I’m on pulumi version 3.67.1 on MacOS. Any suggestions on what might be going on?

are you creating resources inside an

.apply

?

@red-match-15116 No, not here.

Are you seeing different resources change when running

pulumi refresh

?

Are the URNs for these resources that show up for deletion static or dynamic?

@elegant-window-55250 Yes, it appears so.

~ aws:glue/catalogTable:CatalogTable: (update) 🔒
        [id=176915871252:ddt:infutor_reverse_email]
        [urn=urn:pulumi:jde-prod::cc-devops::aws:glue/catalogTable:CatalogTable::inf_reverse_email]
        [provider=urn:pulumi:jde-prod::cc-devops::pulumi:providers:aws::default_5_30_0::119798e2-9163-4d92-9813-9f1e1596a8f8]
        --outputs:--
      + partitionIndices : []

pulumi refresh

actually wants to update all of my Glue Tables with

update     [diff: +partitionIndices]

— but in all cases they appear to be an empty array. It also wants to refresh a bunch of other resources. All look to be minor things on that front.