Race-condition in refetchQueries (cache-first query completes before network-only query)

Intended outcome:

Using refetchQueries on a mutation should guarantee to update the store from the network query result.

Actual outcome:

If the application triggers the same query, that was supposed to be refetched, shortly after the mutation has run (which by default will happen with the cache-first policy), this query will typically hit the cache and complete before the network-only query issued by refetchQueries will finish.

Since the refetch query has a lower requestId its result will not be propagated to the store, see: https://github.com/apollographql/apollo-client/blob/d420cbdbb3e6637ee2e8e8c211e3d28850b93c6e/src/queries/store.ts#L135

How to reproduce the issue:

See: https://github.com/ctavan/react-apollo-error-template/tree/reproduce-refetch-query-race

I think the issue is best explained when looking at the demo app / screenshot:

The steps to reproduce are:

1. Load the page -> List query is run and populates the store with 3 people.
2. Go to “create” which unmounts the List and mounts the Create component. Clicking the button will fire a mutation which adds a new person.
3. The mutation has refetchQueries: ['People'], specified.

• As is visible from the console output which simulates the server behavior, apollo-client starts a network request to refetch the query right after the mutation network request has resolved (and even before the mutate() promise resolves, as indicated by Mutation done). In the example this gets lastRequestId = 4:

• At this point the success callback of the resolved mutation redirects back to the List component which re-triggers the list-query there, lastRequestId = 5:

• Shortly after that, the network request for the refetch resolves and as can be seen from the server output there were actually 4 elements returned.
• However the store still only holds 3 elements for the people query as can be seen from the Apollo client dev toolbar. I assume this is because the cache-first which was executed synchronously resolved before the network-only refetch query and the network-only query was therefore discarded.

My assumption would be that refetchQueries guarantees to sync the current store with the result returned from the server and that intermediate cache-first queries cannot interfere with this.

So, is this a bug or a feature? 😃

[ctavan]

The issue still persists and I believe this issue should not be closed.

[stale[bot]]

This issue has been automatically marked as stale becuase it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions to Apollo Client!