Redis

errorcore instruments the ioredis client library when IO recording is active. Commands are recorded with their name, key target, and duration.

Setup

Initialize errorcore before creating your Redis client:

import errorcore from "errorcore";

errorcore.init({
  service: "session-api",
  transport: { type: "file", path: "./errors.ndjson" },
  allowUnencrypted: true,
});

// create the client after init()
import Redis from "ioredis";
const redis = new Redis(process.env.REDIS_URL);

What gets recorded

Each command in the IO timeline includes:

• command name (GET, SET, HGET, LPUSH, DEL, etc.)
• key or key pattern
• duration in milliseconds
• position in the ordered timeline

 2ms  GET session:abc123
 4ms  HGET user:789 preferences
 9ms  SET session:abc123 (expired)
11ms  Error thrown: session not found

Values

Full values are not captured. Redis values can be large (serialized objects, cached responses, queue payloads) and rarely help explain why an error occurred. The command name and key are usually sufficient to understand the operation sequence.

Pipelines and transactions

Commands inside pipelines and MULTI/EXEC blocks are recorded individually. Each command appears as its own entry in the timeline with its own duration, so you can see whether a specific command within a transaction was the slow one.

Common patterns

Redis capture is most useful for:

• Cache misses: when a missing key triggers a fallback path that fails
• Queue operations: when BRPOP or LPUSH races with other consumers
• Distributed locks: when SET NX coordination does not behave as expected
• Rate limiting: when counter keys expire or do not exist when checked

The timeline shows the order of commands relative to the error, which is often enough to identify the race condition or missing key.