Viewing Logs

Any build or deployment logs emitted to standard output or standard error ( eg. console.log(...)) is captured by Railway to be viewed or searched later.

There are three ways to view logs in Railway.

  • Build/Deploy Panel → Click on a deployment in the dashboard
  • Log Explorer → Click on the Observability tab in the top navigation
  • CLI → Run the railway logs command

Build/Deploy Panel

Logs for a specific deployment can be viewed by clicking on the deployment in dashboard, useful when debugging build or deployment failures.

Logs for a specific deployment

Log Explorer

Logs for the entire environment can be viewed together by clicking the "Observability" button in the top navigation. The Log Explorer is useful for debugging more general problems that may span multiple services.

The log explorer also has additional features like selecting a date range or toggling the visibility of specific columns.

Railway Log Explorer

Command Line

Deployment logs can also be viewed from the command line to quickly check the current status of the latest deployment. Use railway logs to view them.

Viewing logs using the command line interface

Filtering Logs

Railway supports a custom filter syntax that can be used to query logs.

  • <keyword> or "key phrase" → Filter by exact text
  • @key:value → Filter by key/value pair
    • Valid keys are replica, deployment, service, plugin
  • @attribute:value → Filter by custom attribute (see structured logs below)

Any of the above expressions can be combined with boolean operators AND, OR, and - (negation).

Find logs containing "error" for a specific service 👇

Find all 404 errors that are NOT from a specific service 👇

Find logs with a custom attribute 👇

View In Context

Often, when searching for a log, it is useful to see the surrounding logs. To do this, either click the "Timestamp" column, or expand any log and click the "View in Context" button.

Viewing a log in context

Structured Logs

Structured logs are logs emitted in a structured JSON format, useful if you want to attach custom metadata to logs or preserve multi-line logs like stack traces.

Structured logs are best generated with a library for your language. For example, the default Winston. JSON format emits logs in the correct structure by default.


Here are a few examples of structured logs. Note that the entire log must be emitted on a single line to be parsed correctly.

Normalization Strategy

In order to ensure a consistent query format across Railway services, incoming logs are normalized to the above format automatically.

  • Non-structured logs are converted to {"msg":"...","level":"info"}
  • log.message converted to log.msg
  • log.severity converted to log.level
  • log.level defaults to info if missing
  • Levels are lowercased and matched to the closest of debug, info, warn, error

Edit this file on GitHub