net_instaweb::RequestTimingInfo Class Reference

#include "request_timing_info.h"

Public Member Functions
	RequestTimingInfo (Timer *timer, AbstractMutex *mutex)
void	RequestStarted ()
void	ProcessingStarted ()
void	ParsingStarted ()
	This should be called if/when HTML parsing begins.
void	FirstByteReturned ()
	Called when the first byte is sent back to the user.
void	PropertyCacheLookupStarted ()
	This should be called when a PropertyCache lookup is initiated.
void	PropertyCacheLookupFinished ()
	This should be called when a PropertyCache lookup completes.
void	RequestFinished ()
void	FetchStarted ()
void	FetchHeaderReceived ()
void	FetchFinished ()
void	SetHTTPCacheLatencyMs (int64 latency_ms)
void	SetL2HTTPCacheLatencyMs (int64 latency_ms)
int64	GetElapsedMs () const
	Milliseconds since Init.
bool	GetTimeToStartProcessingMs (int64 *elapsed_ms) const
	Milliseconds from request start to processing start.
bool	GetProcessingElapsedMs (int64 *elapsed_ms) const
bool	GetTimeToPropertyCacheLookupStartMs (int64 *elapsed_ms) const
	Milliseconds from request start to pcache lookup start.
bool	GetTimeToPropertyCacheLookupEndMs (int64 *elapsed_ms) const
	Milliseconds from request start to pcache lookup end.
bool	GetHTTPCacheLatencyMs (int64 *latency_ms) const
	HTTP Cache latencies.
bool	GetL2HTTPCacheLatencyMs (int64 *latency_ms) const
bool	GetTimeToStartFetchMs (int64 *elapsed_ms) const
	Milliseconds from request start to fetch start.
bool	GetFetchHeaderLatencyMs (int64 *latency_ms) const
	Milliseconds from fetch start to header received.
bool	GetFetchLatencyMs (int64 *latency_ms) const
	Milliseconds from fetch start to fetch end.
bool	GetTimeToFirstByte (int64 *latency_ms) const
bool	GetTimeToStartParseMs (int64 *elapsed_ms) const
	Milliseconds from request start to parse start.
int64	init_ts_ms () const
int64	start_ts_ms () const

Detailed Description

RequestTimingInfo tracks various event timestamps over the lifetime of a request. The timeline looks (roughly) like the following, with the associated RequestTimingInfo calls.

• Request Received/Context created: Init <queueing delay>="">
• Trigger: RequestStarted <option lookup>="">
• Start Processing: ProcessingStarted
• Lookup Properties?: PropertyCacheLookup*
• Fetch?: Fetch*
• Start parsing?: ParsingStarted
• First byte sent to client: FirstByteReturned.
• Finish: RequestFinished NOTE: This class is thread safe.

Constructor & Destructor Documentation

net_instaweb::RequestTimingInfo::RequestTimingInfo	(	Timer *	timer,
		AbstractMutex *	mutex
	)

Initialize the TimingInfo with the specified Timer. Sets init_ts_ to Timer::NowMs, from which GetElapsedMs is based. NOTE: Timer and mutex are not owned by TimingInfo.

Member Function Documentation

void net_instaweb::RequestTimingInfo::FetchStarted

(

)

Fetch related timing events. Note: Only the first call to FetchStarted will have an effect, subsequent calls are silent no-ops.

Todo:

TODO(gee): Fetch and cache timing is busted for reconstructing resources with multiple inputs.

bool net_instaweb::RequestTimingInfo::GetProcessingElapsedMs

(

int64 *

elapsed_ms

)

const

Milliseconds spent "processing": end time - start time - fetch time.

Todo:

TODO(gee): This naming is somewhat misleading since it is from request start not processing start. Leaving as is for historical reasons, at least for the time being.

bool net_instaweb::RequestTimingInfo::GetTimeToFirstByte

(

int64 *

latency_ms

)

const

Milliseconds from receiving the request (Init) to responding with the first byte of data.

void net_instaweb::RequestTimingInfo::ProcessingStarted ( )

inline

This should be called once the options are available and PSOL can start doing meaningful work.

void net_instaweb::RequestTimingInfo::RequestFinished ( )

inline

Called when the request is finished, i.e. the response has been sent to the client.

void net_instaweb::RequestTimingInfo::RequestStarted

(

)

This should be called when the request "starts", potentially after queuing. It denotes the request "start time", which "elapsed" timing values are relative to.

void net_instaweb::RequestTimingInfo::SetHTTPCacheLatencyMs

(

int64

latency_ms

)

Todo:

TODO(gee): I'd really prefer these to be start/end calls, but the WriteThroughCache design pattern will not allow for this.

---

The documentation for this class was generated from the following file:

• pagespeed/opt/logging/request_timing_info.h