The problem has been resolved. Git changes should now be processed as before.

We have a final solution in place and are now manually processing all repositories to make sure we didn't miss any exercise updates. After that's done, the webhooks should operate as normal.

We are working on a solution where a dedicated machine with low latency storage handles all git operations and syncs the result to a network drive used by our other machines. Initial benchmarks indicate a 700x performance improvement over the current situation.

We finalized a pull of all repositories, but noticed that triggering the webhook large exercise repositories is still slow. We hope to have a solution in place early next week.

After the migration, there was an issue where triggering the webhooks for exercise repository updates failed. This has since been resolved.

As a side effect of the migration, the first time the webhook for a repository is triggered, this can take a (very) long time. Subsequent webhook triggers should be acceptably fast. We are now manually triggering the webhooks for all repositories to minimize user impact.