README
¶
zlg (zerolog-gcp)
zlg is a (hopefully) straightforward LevelWriter for using zerolog with Google Cloud Operations Logging, which used to be named Stackdriver.
Some notable features:
- The first log written to Cloud Logging is a slow, blocking write to confirm connectivity + permissions, but all subsequent writes are non-blocking.
- Flushes all logs before a
.Fatal()or.Panic() - Ensure that all zlg-created loggers are flushed before program exit with
defer zlg.Flush() - Overridable Cloud Monitoring levels (zerolog -> Cloud Logging):
- Trace -> Default
- Warn -> Warning
- Cloud Logging's Alert and Emergency levels are not used.
Getting Started
The usual cases
Logging only to Stackdriver:
import zlg "github.com/mark-ignacio/zerolog-gcp"
// [...]
gcpWriter, err := zlg.NewCloudLoggingWriter(ctx, projectID, logID, zlg.CloudLoggingOptions{})
if err != nil {
log.Panic().Err(err).Msg("could not create a CloudLoggingWriter")
}
log.Logger = log.Output(gcpWriter)
For non-GCP-hosted situations, you can log to both stdout/stderr and GCP without much additional fuss.
gcpWriter, err := zlg.NewCloudLoggingWriter(ctx, projectID, logID, zlg.CloudLoggingOptions{})
if err != nil {
log.Panic().Err(err).Msg("could not create a CloudLoggingWriter")
}
log.Logger = log.Output(zerolog.MultiLevelWriter(
zerolog.NewConsoleWriter(),
gcpWriter,
))
To ensure that the last asynchronous logs are delivered to Cloud Logging, zlg keeps a reference to all logging.Logger structs that zlg itself creates. If memory leaks of loggers are a concern, consider specifying providing a logger client via zlg.CloudLoggingOptions instead.
gcpWriter, err := zlg.NewCloudLoggingWriter(ctx, projectID, logID, zlg.CloudLoggingOptions{})
if err != nil {
log.Panic().Err(err).Msg("could not create a CloudLoggingWriter")
}
defer zlg.Flush()
doEverythingElse()
More advanced usage involves a non-empty zlg.CloudLoggingOptions, which allows for log level and GCP client customization.
Documentation
¶
Index ¶
Constants ¶
This section is empty.
Variables ¶
var DefaultSeverityMap = map[zerolog.Level]logging.Severity{ zerolog.DebugLevel: logging.Debug, zerolog.InfoLevel: logging.Info, zerolog.WarnLevel: logging.Warning, zerolog.ErrorLevel: logging.Error, zerolog.PanicLevel: logging.Critical, zerolog.FatalLevel: logging.Critical, }
DefaultSeverityMap contains the default zerolog.Level -> logging.Severity mappings.
Functions ¶
func Flush ¶ added in v0.2.0
func Flush() []error
Flush blocks while flushing all loggers this module created.
func NewCloudLoggingWriter ¶
func NewCloudLoggingWriter(ctx context.Context, projectID, logID string, opts CloudLoggingOptions) (writer zerolog.LevelWriter, err error)
NewCloudLoggingWriter creates a LevelWriter that logs only to GCP Cloud Logging using non-blocking calls.
Types ¶
type CloudLoggingOptions ¶
type CloudLoggingOptions struct {
// Specify this to override DefaultSeverityMap.
SeverityMap map[zerolog.Level]logging.Severity
// Used during *logging.Client construction.
ClientOptions []option.ClientOption
// Used during *logging.Client construction.
ClientOnError func(error)
// Specify this to override the default of constructing a *logging.Logger on the caller's behalf.
Logger *logging.Logger
// Used during GCP Logger construction.
LoggerOptions []logging.LoggerOption
}
CloudLoggingOptions specifies some optional configuration.
Source Files