OperationStore index appears to fail with nil comparsion about graphql-ruby HOT 10 CLOSED

Hey, sorry for the trouble and thanks for reporting this! "Archiving" was a later addition to the OperationStore, but I certainly intended to retain backwards compatibility for systems that didn't migrate or hadn't migrated yet. I'll investigate this soon, and if you want to try migrating your database in the meantime, you can find the migration notes here: https://graphql-ruby.org/operation_store/active_record_backend.html#database-update

from graphql-ruby.

Hmm - we did the migration referenced here a while ago. I can confirm that both of those columns are in the DB. But the error appears to be complaining about archived_references_count, which from what I can tell is actually just computed in a join.

So it sounds like there's nothing that jumps out obviously as a step we missed, which makes me wonder if maybe we have something else slightly unexpected in our system.

from graphql-ruby.

Yeah... very weird. Maybe that SUM(...) is somehow returning null instead of 0. What database are you using?

from graphql-ruby.

We're using Postgres 12.18! Sorry, should have included that earlier.

from graphql-ruby.

No worries, thanks. I was able to replicate this issue when GraphQLIndexReference records are missing from the database. I'm not sure how that could happen, but it's easy enough to recreate them in a Rails console:

MySchema.operation_store.reindex

I expect if you run that in a console, this view will be fixed for you.

In any case, I also released graphql-pro 1.27.2 with more graceful handling for this case, printing a message suggesting that same fix.

Could you try that fix, or try the new version, and let me know how it goes for you?

from graphql-ruby.

Hmm - MySchema.operation_store.reindex seemed to fix the backend, but that command seems to run forever. Am I missing something, or does the reindex method never actually increment current_page? It seems odd that this is an eternal loop for me, given that at least iterations should increment to 1000 and stop?

from graphql-ruby.

Hmm, I spoke prematurely - it no longer has the same error, but the index is pretty obviously incomplete. :-/

from graphql-ruby.

Derp... so sorry about that. I just released graphql-pro v1.27.3 with a fixed (and properly tested...) multi-page .reindex behavior. Could you try again on the new version?

from graphql-ruby.

Ah ha! That got us back in shape. Thanks a ton, @rmosolgo !

from graphql-ruby.

Glad to hear it. Sorry for the trouble and thanks for working through it with me!

from graphql-ruby.