process.exit(1) on Any Single User's API Failure Destroys Entire Sync Cycle

[rishab11250]

Severity

Critical — Makes the entire sync pipeline unreliable

Location

scripts/sync-leaderboard.js:14-15

Description

fetchData() terminates the entire Node process if any single user's API call fails:

async function fetchData(url) {
  try {
    const res = await axios.get(url);
    return {
      easySolved: res.data.easySolved || 0,
      mediumSolved: res.data.mediumSolved || 0,
      hardSolved: res.data.hardSolved || 0
    };
  } catch (err) {
    console.error("API failed to respond: ", err.message);
    process.exit(1);   // kills everything
  }
}

With 10+ users hitting a third-party API, transient failures are guaranteed. When
one user fails:

• All previously fetched data (other users) is lost
• CI iteration running at 300s intervals is wasted
• With continue-on-error: true, the failure is invisible
• No Axios timeout (default: 0 = infinite) means a hang kills the entire 6-hour run

Why It Matters

• The sync script generates all leaderboard data — it's the most critical function
• Third-party APIs are inherently unreliable
• A single rate-limited user means zero data for everyone that cycle
• No partial-results fallback

Fix

async function fetchData(url) {
  try {
    const res = await axios.get(url, { timeout: 15000 });
    return {
      easySolved: res.data.easySolved || 0,
      mediumSolved: res.data.mediumSolved || 0,
      hardSolved: res.data.hardSolved || 0
    };
  } catch (err) {
    console.error(`API failed for ${url}: ${err.message}`);
    return null;  // signal failure without killing process
  }
}

// In the loop:
if (data === null) {
  console.log(`Skipping ${user.name} due to API error`);
  continue;
}

[rishab11250]

I want to work on this issue

[jagdish-15]

Thanks!

We've previously migrated to our own hosted API, which reduces some of the concerns around third-party rate limits and availability. However, the underlying issue remains valid: a failure while fetching data for a single user should not abort the entire sync cycle.

Assigning this issue to @rishab11250 since they opened the issue and expressed interest in working on it first.