== Evergreen/OpenSRF log levels ==


OpenSRF supports 5 different logging levels:  ERROR, WARN, INFO, DEBUG, and INTERNAL.  They are controlled in the "bootstrap" configuration file, usually found at /openils/conf/opensrf_core.xml.  Log levels are configured with a number, 1 to 5, where  1 equals ERROR and 5 equals INTERNAL.  There are 4 classifications of services across which the configured log level is shared.  These include Apache (gateway), the OpenSRF routers, OpenSRF services, and srfsh.  In a production environment, the following log levels are recommended for each service.

  * Apache Gateway -- INFO
  * OpenSRF Routers -- WARN
  * OpenSRF services -- INFO
  * srfsh -- INFO (or DEBUG)


== Description of log levels ==

**ERROR**:  Error logging only shows system and application errors.  These errors typically indicate an application-level exception and are non-fatal to the system as a whole.  However, if systemic problems arise, they will show up in this log as well.

**WARN**:  There isn't a lot of content that logs at this level.  The most useful data that is logged as a WARNing are instances of exhausted C service drone processes.  A certain amount of this is expected, but many occurrences of this on a daily basis is an indication of insufficient resources or resource starvation.

2008-01-14 11:32:08 app123 open-ils.cstore: [WARN:877:osrf_prefork.c:515:] We have no children available - waiting for one to show up...

**INFO**:  Info logging is used to indicate application-level activity.  Info logs provide details on inbound API requests, execution of custom policy scripts (circulation / holds), and high-level flow of execution.  Most problems can be diagnosed at INFO level.  Info logging also includes API call timing data.  For example:

2009-08-26 00:00:00 app123 open-ils.cstore: [INFO:9762:osrf_stack.c:116:] Message processing duration 0.001547

The Message processing duration indicates the amount of time between receiving a request and sending the response.  This data is logged for all OpenSRF services at all levels.  This data is also logged for OpenSRF client applications, which means top-level HTTP requests which are translated into OpenSRF calls are logged with the call duration as well.  This results in a system that produces timing data for each request to each OpenSRF component for ease of identifying bottlenecks in software execution.

**DEBUG**: Debug logs should only be used for development or debugging purposes.  They include a high volume of log data which is generally only useful for developers.

**INTERNAL**:  This is debug logging on stereoids.  It should never be used.