Default JSON Logging vs. Plain console.log: Which is Better for Clustered Deployment Log Rotation?

Question

The primary goal is to reliably diagnose deployment failures in a Node.js application that uses KrakenJS’s kraken-logger. In a clustered or PM2 environment, each worker writes to its own log file, and the built‑in rotation option only limits the size of a single process’s file.
When logs are rotated independently, stale files can accumulate and duplicate entries obscure the failure timeline. Conversely, plain console.log writes unstructured text to stdout, which is easier to view locally but harder for automated ingestion.
Because there is no out‑of‑the‑box mechanism to coordinate rotation or consolidate logs across workers, the team must decide how to balance readability, aggregation, and file management.
How can we coordinate log rotation across PM2 cluster workers when using kraken-logger?What is the recommended approach to consolidate logs from multiple processes for a single aggregation pipeline?How does the trade‑off between JSON logging and plain console output affect parsing by services like Loki?

Wasi · Answer

Direct answer
Do not rely on kraken-logger’s built-in rotation to coordinate across PM2 cluster workers. That option typically opens a file handle per process, so each worker rotates its own file independently and there is no shared lock or global size cap. Write structured JSON lines to stdout, let PM2 capture the stream, and delegate rotation and consolidation to pm2-logrotate plus a log shipper such as Promtail or Fluent Bit feeding Loki.
Likely explanation vs. confirmed facts

Likely explanation: kraken-logger’s per-worker file rotation cannot enforce one global limit because PM2 cluster workers are separate processes with independent file descriptors. Plain console.log avoids file rotation entirely but ships unstructured text to stdout.
Confirmed facts: PM2 cluster mode forks independent workers; each worker’s stdout/stderr is captured by PM2 unless redirected. Loki ingests JSON lines natively with a json pipeline stage, while plain text requires regex extraction. Multiple writers to a single file can interleave lines unless concurrent-safe writes are guaranteed.

Steps for this case

Configure kraken-logger to emit one JSON object per line to stdout, including level, timestamp, requestId, and message. Escape multiline stack traces into a single string so Loki does not split them.
Start the app with PM2 in cluster mode: pm2 start app.js -i max --name api. PM2 writes worker stdout to its own log files under ~/.pm2/logs.
Install and configure PM2 log rotation for the captured stream:
pm2 install pm2-logrotate
pm2 set pm2-logrotate:max_size 10M
pm2 set pm2-logrotate:retain 7
pm2 set pm2-logrotate:compress true

Point Promtail or Fluent Bit at the PM2 log directory, parse each line as JSON, and label by instance, level, and requestId before sending to Loki.
Verify with pm2 logs api --lines 20 and a Loki query such as {app="api"} | json | level = "error".

JSON vs. plain console.log for Loki

OutputLoki parsingOperational note

JSON linesAutomatic field extraction; filter by level, requestId without regex.Slightly larger lines; multiline content must be escaped into one line.
Plain console.logRequires regex or pattern stages; higher ingestion CPU and fragile field extraction.Easier to read locally during manual debugging.

Assumption: kraken-logger supports streaming JSON to stdout in your installed version; rotation defaults and API vary by version, so verify package.json before depending on file-based rotation.
Missing diagnostic detail: What version of kraken-logger is installed, and is it configured to write to stdout or to per-worker files? This determines whether PM2 can capture the stream directly or whether you need a file tailer.

Default JSON Logging vs. Plain console.log: Which is Better for Clustered Deployment Log Rotation?

1 answer

Direct answer

Likely explanation vs. confirmed facts

Steps for this case

JSON vs. plain console.log for Loki

Your perspective belongs here.

1 question comment

Coordination Strategy

Implementation

Verification

Output	Loki parsing	Operational note
JSON lines	Automatic field extraction; filter by `level`, `requestId` without regex.	Slightly larger lines; multiline content must be escaped into one line.
Plain `console.log`	Requires regex or pattern stages; higher ingestion CPU and fragile field extraction.	Easier to read locally during manual debugging.