Logging

We mainly use 2 loggers that are defined in the cells/common/log package and used throughout the backend:

• A syslog logger is used to log technical information.
• An audit logger only logs precise information and is used to monitor the application from a business point of view. Audit logs are among others used for GRDP compliance. WARNING: you should not modify nor remove any of the audit log.

Pydio Cells relies on the zap logging framework that manages 5 levels of information:

• Debug
• Info
• Warn
• Error
• Fatal

And 2 types of output format:

• JSON
• Console like with colors.

Depending on the launching mode (see below), the logs are redirected to:

• Standard out
• Rolling files
• A gRPC service that streams each message in JSON format to a log service that store them in a BoltDB and index them with Bleve.

The logs can then be accessed within the admin dashboard.

Currently, the application can be launched in one of the three following configuration:

• Debug: all log messages are redirected to standard out in a console friendly format
• Info (this is the default mode): same as debug mode but all message with debug level are dropped
• Production:
  • Debug message are dropped
  • Message are formatted as JSON
  • Messages are redirected to the standard output, some rolling files and streamed the gRPC service that stores and indexes the messages.

Guidelines

• When enriching the log with zap fields, always use constants from common/zapfields.go as keys, otherwise the logs won't be usable for analysis
• Pydio provides some shortcuts to easily generate zap fields for commons pydio objects. Typically when logging info about a tree.Node, you should write: ...Debug("A message", node.Zap())
• Depending on the level of details required, you can either use Zap() that serializes all accessible information or ZapPath(), ZapUuid() or ZapId() (depending on the relevant info) that simply calls zap.String() with relevant pydio zap id (as seen in common/zapfields.go) and corresponding value.