Timeouts when fetching commits using changed_files, additions and/or deletions

GitHub Ecosystem
#1

Hello!

I’ve been exploring the graph v4 api to fetch commits of different repos.
In some cases, a timeout will be thrown with a message “Something, somewhere, has gone horribly, horribly wrong.” when executing a somewhat simple querries.

It only happens when commits in the page have many changes, like 50k or somewhat similarly. I believe the count, which is stated to be an int, is made on the server side after calling the query, thus resulting in a timeout in cases where the query is very simple.

Following is an example query that creates an timeout:

{
  repository(owner:"eclipse", name: "rdf4j") {
    defaultBranchRef {
        target {
            ... on Commit {
                id
                history(first: 4, after: "a12d70d4e391413f5037e65824f4748a72504a28 94") {
                    pageInfo {
                        hasNextPage
                        startCursor
                        endCursor
                    }
                    nodes {
                        changedFiles
                    }
                }
            }
        }
    }
  }
}
#2

Hello @ProgHaj and welcome to the community.

Yes, over and above the rate limit restrictions on GraphQL queries, there is a timeout on queries that take too long to return. Accessing git data is one way that, as you noticed, even simple queries can run afoul of that timeout.

Let us know if you have more questions.