Free Sys-Arch-301 sample questions

Correct answer: The risk is infinite recursion; implement a static boolean variable in a handler class to control trigger execution state.

The primary risk here is infinite recursion where the Account update triggers a Contact update, which triggers an Account update, creating a loop until the governor limit is hit. The standard mitigation pattern is to use a static boolean variable in a handler class to track whether the trigger has already run in the current transaction context.

Correct answer: Batch Apex with `Database.AllowsCallouts`

Batch Apex is designed for processing large volumes of records (thousands). To make callouts, the class must implement the `Database.AllowsCallouts` interface. Future methods cannot process large record sets easily, and Queueable is better for chaining jobs but less ideal for iterating over query locators of thousands of records.

Correct answer: The query will fail with a System.QueryException due to low selectivity.

Even though `LastName` is indexed, returning 2 million out of 5 million records exceeds the selectivity threshold (usually 10% of records for standard indexes). The query optimizer will determine that the filter is not selective enough and will likely perform a full table scan, resulting in a timeout or exception.

Correct answer: A Custom Event with `bubbles: true` and `composed: true`

To communicate from a deep child to a parent across shadow DOM boundaries, dispatching a Custom Event with `bubbles: true` (to bubble up) and `composed: true` (to cross shadow DOM boundaries) is the standard pattern. LMS is typically used for communication between unrelated components or across the DOM tree laterally.

Correct answer: respond(HttpRequest req)

The `HttpCalloutMock` interface requires the implementation of the `respond(HttpRequest req)` method, which returns an `HttpResponse` object.

Correct answer: Distribute the child contacts across multiple dummy accounts (e.g., Unassigned_1, Unassigned_2) to keep child counts below 10,000 per parent.

Data skew occurs when a single parent record has more than 10,000 child records. The best practice is to distribute these children across multiple parent records to prevent lock contention and sharing calculation performance issues.