cache.md

Problems Cache Layer

The Problems Cache uses the Decorator Pattern to intercept and cache the heavy MongoDB queries executed by the repositories.

File Location: api/src/cache/problems/

1. ProblemCache (problem.cache.ts)

Because coding problems rarely change, they are aggressively cached.

Strategies:

• findMany: Caches the paginated grid view.
  • Key: problems:list:{limit}:{offset}:{filters}
  • TTL: 300 seconds (5 minutes). This is short enough that new problems appear quickly, but long enough to absorb traffic spikes.
• findBySlug: Caches the individual problem details.
  • Key: problem:slug:{slug}
  • TTL: 86400 seconds (24 hours). The details of a problem almost never change.
• Invalidation: When create or update is called, the cache immediately calls redis.del() on the specific slug key and wildcard flushes the problems:list:* keys so the grid instantly updates.

2. ProblemTestCache (problem-test.cache.ts)

Test cases are fetched constantly by the backend Judge worker during evaluation.

Strategy:

• findAllByProblem: Caches both public and hidden test cases together.
  • Key: problem-tests:{problemId}
  • TTL: 86400 seconds (24 hours).
• Worker Optimization: By caching this, the Judge worker (SubmissionEvaluator) can fetch all 100 hidden test cases from Redis memory in milliseconds rather than hitting MongoDB on every code submission!