Documentation ¶
Overview ¶
Package sdlogger is a logger that formats entries as JSON understood by Stackdriver log collectors.
The JSON format is defined by the Google Cloud Logging plugin for fluentd, which is mainly used by Google Kubernetes Engine + Stackdriver stack. See https://github.com/GoogleCloudPlatform/fluent-plugin-google-cloud
Though not properly documented, GAE environment also understands subset of this format.
Index ¶
Constants ¶
This section is empty.
Variables ¶
This section is empty.
Functions ¶
func Factory ¶
func Factory(w LogEntryWriter, prototype LogEntry, mut LogEntryMutator) func(context.Context) logging.Logger
Factory returns a factory of logger.Logger instances that log to the given LogEntryWriter (usually &Sink{...}) using the LogEntry (if any) as a prototype for log entries.
For each log message, makes a copy of 'prototype', overwrites its Severity, Message, Time and Fields (keeping other fields as they are), calls 'mut' callback (which is allowed to mutate the log entry), and writes the result to LogEntryWriter. This allows to "prepopulate" fields like TraceID or OperationID in all log entries emitted by logging.Logger.
Such factory can be installed in the context via logging.SetFactory.
Types ¶
type CloudErrorsSink ¶
type CloudErrorsSink struct { Client *errorreporting.Client Request *http.Request Out LogEntryWriter }
CloudErrorsSink wraps a LogEntryWriter and an errorreporting.Client.
func (*CloudErrorsSink) Write ¶
func (s *CloudErrorsSink) Write(l *LogEntry)
Write appends a log entry to the given writer and reports the message to the cloud error reporting for the Error severity.
type LogEntry ¶
type LogEntry struct { // Severity is a string denoting the logging level of the entry. Severity Severity `json:"severity"` // Message is a single line human readable string of the log message. Message string `json:"message,omitempty"` // Timestamp is the unix timestamp of when the entry was produced. Timestamp Timestamp `json:"timestamp,omitempty"` // TraceID is 32-byte hex string with Stackdriver trace ID. TraceID string `json:"logging.googleapis.com/trace,omitempty"` // TraceSampled is true if this trace will be uploaded to Stackdriver. TraceSampled bool `json:"logging.googleapis.com/trace_sampled,omitempty"` // SpanID is a 16-byte hex string with Stackdriver span ID. SpanID string `json:"logging.googleapis.com/spanId,omitempty"` // Operation is used to group log lines from a single request together. Operation *Operation `json:"logging.googleapis.com/operation,omitempty"` // RequestInfo is information about the handled HTTP request. RequestInfo *RequestInfo `json:"httpRequest,omitempty"` // Fields are extra structured data that may be tagged in the log entry. Fields logging.Fields `json:"fields,omitempty"` }
LogEntry defines a structure of JSON log entries recognized by google-fluentd plugin on GKE.
See https://cloud.google.com/logging/docs/agent/configuration#process-payload for a list of possible magical fields. All other fields are exposed as part of jsonPayload in Stackdriver logs.
type LogEntryMutator ¶
LogEntryMutator may mutate log entries before they are written.
type LogEntryWriter ¶
type LogEntryWriter interface { // Write appends a log entry to the output. Write(*LogEntry) }
LogEntryWriter knows how to write LogEntries to some output.
type Operation ¶
type Operation struct {
ID string `json:"id"`
}
Operation is used to group log lines from a single request together.
type RequestInfo ¶
type RequestInfo struct { Method string `json:"requestMethod"` // e.g. "GET" URL string `json:"requestUrl"` // e.g. "http://host/path" Status int `json:"status"` // e.g. 200 RequestSize string `json:"requestSize"` // e.g. "123123" ResponseSize string `json:"responseSize"` // e.g. "2324" UserAgent string `json:"userAgent"` // e.g. "Mozilla/4.0 ..." RemoteIP string `json:"remoteIp"` // e.g. "192.168.1.1" Latency string `json:"latency"` // e.g. "3.5s" }
RequestInfo contains information about handled HTTP request.
See following definition for all fields we can possibly expose here: https://cloud.google.com/logging/docs/reference/v2/rest/v2/LogEntry#HttpRequest
type Severity ¶
type Severity string
Severity is a string denoting the logging level of the entry.
Must be one of the predefined constants, otherwise Stackdriver may not recognize it.
const ( UnknownSeverity Severity = "" DebugSeverity Severity = "DEBUG" InfoSeverity Severity = "INFO" WarningSeverity Severity = "WARNING" ErrorSeverity Severity = "ERROR" )
All allowed severities. Using other values results in undefined behavior.
func LevelToSeverity ¶
LevelToSeverity converts logging level to the corresponding severity.
type SeverityTracker ¶
type SeverityTracker struct { Out LogEntryWriter // contains filtered or unexported fields }
SeverityTracker wraps LogEntryWriter and observes severity of messages there.
func (*SeverityTracker) MaxSeverity ¶
func (s *SeverityTracker) MaxSeverity() Severity
MaxSeverity returns maximum severity observed thus far or "".
func (*SeverityTracker) Write ¶
func (s *SeverityTracker) Write(l *LogEntry)
Write is part of LogEntryWriter interface.
type Sink ¶
Sink takes care of JSON-serializing log entries and synchronizing writes to an io.Writer (usually stdout or stderr).
Implements LogEntryWriter.
Sink can either be used directly for fine-grain control of what is getting logged, or via a logging.Logger for interoperability with most of LUCI code.
There should be at most one Sink instance assigned to a given io.Writer, shared by all Logger instances writing to it. Violating this requirement may cause malformed log lines.