Symfony Monolog fingers_crossed: keeping pre-error debug records from a failed deployment run

To guarantee that failed deployment commands leave a usable trace without flooding your production logs, you should separate your CLI logging configuration from your HTTP configuration. The fingers_crossed handler is designed for request-response cycles; using it for long-running deployment scripts creates a risk where logs are discarded if the process exits without hitting the action_level.

The Recommended Solution: Scoped CLI Handlers

The most reliable way to handle this in Symfony is to define a separate handler specifically for the console. This allows you to use a standard stream handler (which writes immediately) or a fingers_crossed handler with a much lower action_level for CLI commands.

In your config/packages/monolog.yaml, you can use the channels configuration to isolate console output:

# config/packages/monolog.yaml
monolog:
    handlers:
        # Standard production HTTP handler
        main:
            type: fingers_crossed
            action_level: error
            handler: nested
        nested:
            type: stream
            path: "%kernel.logs_dir%/%kernel.environment%.log"
            level: debug

        # Dedicated CLI handler
        console:
            type: stream
            path: "%kernel.logs_dir%/deploy.log"
            level: info # Capture everything INFO and above immediately
            channels: ["console"]

Addressing Your Specific Constraints

Does passthru_level interact with buffer_size?

Yes. buffer_size defines the maximum number of records kept in memory. If your deployment script generates 1,000 debug lines but your buffer_size is 100, the first 900 lines are discarded before the action_level is ever reached. If a failure then triggers the flush, you only see the last 100 records. passthru_level does not prevent this; it only ensures that records at or above that level are written regardless of whether the action_level was triggered.

Forcing a buffer flush in Console Commands

There is no native monolog.yaml configuration to "force flush on exit" if the activation level wasn't met. To achieve this, you would need to manually inject the logger into your command and call the handler's close/flush method, or more simply, avoid fingers_crossed for CLI tasks entirely.

Summary of Behavior

Handler Type	Behavior on Success	Behavior on Failure	Risk
fingers_crossed	Logs discarded	Buffer flushed to disk	Buffer overflow (loss of early logs)
stream	All logs written	All logs written	Log file bloat

Diagnostic Detail Needed: Are your deployment failures resulting in PHP Fatal Errors (Segmentation faults/Out of Memory) or Symfony-caught Exceptions? If the process crashes at the PHP engine level, Monolog's shutdown functions may not execute, meaning no buffer will ever be flushed regardless of configuration.