Container Logs

Cycle automatically aggregates container logs from stdout from all instances. Logs are timestamped and queryable.

How Container Logs Work

Unlike the instance console, which only stores 128KB of output between instance restarts, container logs are intended to be stored long term and searchable/queryable. Logs can be searched via raw text (fast), or using regex (slower, but more powerful).

Cycle will automatically aggregate 72 hours of logs from an instance's stdout on the local server the instance is running on. The maximum log recording is 3,000 lines per minute per instance, and any lines above 3,000 in that timeframe will be discarded.

Logs are written to a server's base volume. If there are many containers running on a server, the base volume size may need to be increased to handle logging. The logs are initially cached in memory, then written to disk every minute.

Duplicate Log Lines

Duplicate log lines are compressed, and the final duplicate is replaced with a line telling how many duplicate lines were skipped.

Query Container Logs on Cycle

To query a container's logs, follow one of our interface-specific docs below:

• Portal
• API