This packetizer is based on TPacketizer but uses different load-balancing algorithms and data structures.

Two main improvements in the load-balancing strategy:

First one was to change the order in which the files are assigned to the computing nodes in such a way that network transfers are evenly distributed in the query time. Transfer of the remote files was often becoming a bottleneck at the end of a query.
The other improvement is the use of time-based packet size. We measure the processing rate of all the nodes and calculate the packet size, so that it takes certain amount of time. In this way packetizer prevents the situation where the query can't finish because of one slow node.

The data structures: TFileStat, TFileNode and TSlaveStat are enriched + changed and TFileNode::Compare method is changed.

Definition at line 50 of file TPacketizerAdaptive.h.

Public Member Functions
	TPacketizerAdaptive (TDSet dset, TList slaves, Long64_t first, Long64_t num, TList input, TProofProgressStatus st)

virtual	~TPacketizerAdaptive ()
	Destructor. More...

Int_t	AddProcessed (TSlave sl, TProofProgressStatus st, Double_t latency, TList **listOfMissingFiles=0)
	To be used by GetNextPacket but also in reaction to kPROOF_STOPPROCESS message (when the worker was asked to stop processing during a packet). More...

Int_t	GetEstEntriesProcessed (Float_t, Long64_t &ent, Long64_t &bytes, Long64_t &calls)
	Get estimation for the number of processed entries and bytes read at time t, based on the numbers already processed and the latests worker measured speeds. More...

Float_t	GetCurrentRate (Bool_t &all)
	Get Estimation of the current rate; just summing the current rates of the active workers. More...

Int_t	CalculatePacketSize (TObject *slstat, Long64_t cachesz, Int_t learnent)
	The result depends on the fStrategy. More...

TDSetElement *	GetNextPacket (TSlave sl, TMessage r)
	Get next packet; A meaningfull difference to TPacketizer is the fact that this packetizer, for each worker, tries to predict whether the worker will finish processing it's local files before the end of the query. More...

void	MarkBad (TSlave s, TProofProgressStatus status, TList **missingFiles)
	This method can be called at any time during processing as an effect of handling kPROOF_STOPPROCESS If the output list from this worker is going to be sent back to the master, the 'status' includes the number of entries processed by the slave. More...

Int_t	GetActiveWorkers ()
	Return the number of workers still processing. More...

Public Member Functions inherited from TVirtualPacketizer
virtual	~TVirtualPacketizer ()
	Destructor. More...

virtual Int_t	AssignWork (TDSet *, Long64_t, Long64_t)

Bool_t	IsValid () const

Long64_t	GetEntriesProcessed () const

Long64_t	GetTotalEntries () const

virtual void	SetInitTime ()
	Set the initialization time. More...

virtual void	StopProcess (Bool_t abort, Bool_t stoptimer=kFALSE)
	Stop process. More...

TList *	GetFailedPackets ()

void	SetFailedPackets (TList *list)

virtual Int_t	AddWorkers (TList *workers)
	Adds new workers. More...

Long64_t	GetBytesRead () const

Long64_t	GetReadCalls () const

Double_t	GetCumProcTime () const

Float_t	GetInitTime () const

Float_t	GetProcTime () const

TNtuple *	GetProgressPerf (Bool_t steal=kFALSE)

TList *	GetConfigParams (Bool_t steal=kFALSE)

TProofProgressStatus *	GetStatus ()

void	SetProgressStatus (TProofProgressStatus *st)

void	SetTotalEntries (Long64_t ent)

TMap *	GetSlaveStats () const

Public Member Functions inherited from TObject
	TObject ()

	TObject (const TObject &object)
	TObject copy ctor. More...

TObject &	operator= (const TObject &rhs)
	TObject assignment operator. More...

virtual	~TObject ()
	TObject destructor. More...

virtual void	AppendPad (Option_t *option="")
	Append graphics object to current pad. More...

virtual void	Browse (TBrowser *b)
	Browse object. May be overridden for another default action. More...

virtual const char *	ClassName () const
	Returns name of class to which the object belongs. More...

virtual void	Clear (Option_t *="")

virtual TObject *	Clone (const char *newname="") const
	Make a clone of an object using the Streamer facility. More...

virtual Int_t	Compare (const TObject *obj) const
	Compare abstract method. More...

virtual void	Copy (TObject &object) const
	Copy this to obj. More...

virtual void	Delete (Option_t *option="")
	Delete this object. More...

virtual Int_t	DistancetoPrimitive (Int_t px, Int_t py)
	Computes distance from point (px,py) to the object. More...

virtual void	Draw (Option_t *option="")
	Default Draw method for all objects. More...

virtual void	DrawClass () const
	Draw class inheritance tree of the class to which this object belongs. More...

virtual TObject *	DrawClone (Option_t *option="") const
	Draw a clone of this object in the current pad. More...

virtual void	Dump () const
	Dump contents of object on stdout. More...

virtual void	Execute (const char method, const char params, Int_t *error=0)
	Execute method on this object with the given parameter string, e.g. More...

virtual void	Execute (TMethod method, TObjArray params, Int_t *error=0)
	Execute method on this object with parameters stored in the TObjArray. More...

virtual void	ExecuteEvent (Int_t event, Int_t px, Int_t py)
	Execute action corresponding to an event at (px,py). More...

virtual TObject *	FindObject (const char *name) const
	Must be redefined in derived classes. More...

virtual TObject *	FindObject (const TObject *obj) const
	Must be redefined in derived classes. More...

virtual Option_t *	GetDrawOption () const
	Get option used by the graphics system to draw this object. More...

virtual UInt_t	GetUniqueID () const
	Return the unique object id. More...

virtual const char *	GetName () const
	Returns name of object. More...

virtual const char *	GetIconName () const
	Returns mime type name of object. More...

virtual Option_t *	GetOption () const

virtual char *	GetObjectInfo (Int_t px, Int_t py) const
	Returns string containing info about the object at position (px,py). More...

virtual const char *	GetTitle () const
	Returns title of object. More...

virtual ULong_t	Hash () const
	Return hash value for this object. More...

virtual Bool_t	InheritsFrom (const char *classname) const
	Returns kTRUE if object inherits from class "classname". More...

virtual Bool_t	InheritsFrom (const TClass *cl) const
	Returns kTRUE if object inherits from TClass cl. More...

virtual void	Inspect () const
	Dump contents of this object in a graphics canvas. More...

virtual Bool_t	IsFolder () const
	Returns kTRUE in case object contains browsable objects (like containers or lists of other objects). More...

virtual Bool_t	IsEqual (const TObject *obj) const
	Default equal comparison (objects are equal if they have the same address in memory). More...

virtual Bool_t	IsSortable () const

Bool_t	IsOnHeap () const

Bool_t	IsZombie () const

virtual Bool_t	Notify ()
	This method must be overridden to handle object notification. More...

virtual void	ls (Option_t *option="") const
	The ls function lists the contents of a class on stdout. More...

virtual void	Paint (Option_t *option="")
	This method must be overridden if a class wants to paint itself. More...

virtual void	Pop ()
	Pop on object drawn in a pad to the top of the display list. More...

virtual void	Print (Option_t *option="") const
	This method must be overridden when a class wants to print itself. More...

virtual Int_t	Read (const char *name)
	Read contents of object with specified name from the current directory. More...

virtual void	RecursiveRemove (TObject *obj)
	Recursively remove this object from a list. More...

virtual void	SaveAs (const char filename="", Option_t option="") const
	Save this object in the file specified by filename. More...

virtual void	SavePrimitive (std::ostream &out, Option_t *option="")
	Save a primitive as a C++ statement(s) on output stream "out". More...

virtual void	SetDrawOption (Option_t *option="")
	Set drawing option for object. More...

virtual void	SetUniqueID (UInt_t uid)
	Set the unique object id. More...

virtual void	UseCurrentStyle ()
	Set current style settings in this object This function is called when either TCanvas::UseCurrentStyle or TROOT::ForceStyle have been invoked. More...

virtual Int_t	Write (const char *name=0, Int_t option=0, Int_t bufsize=0)
	Write this object to the current directory. More...

virtual Int_t	Write (const char *name=0, Int_t option=0, Int_t bufsize=0) const
	Write this object to the current directory. More...

void *	operator new (size_t sz)

void *	operator new[] (size_t sz)

void *	operator new (size_t sz, void *vp)

void *	operator new[] (size_t sz, void *vp)

void	operator delete (void *ptr)
	Operator delete. More...

void	operator delete[] (void *ptr)
	Operator delete []. More...

void	SetBit (UInt_t f, Bool_t set)
	Set or unset the user status bits as specified in f. More...

void	SetBit (UInt_t f)

void	ResetBit (UInt_t f)

Bool_t	TestBit (UInt_t f) const

Int_t	TestBits (UInt_t f) const

void	InvertBit (UInt_t f)

virtual void	Info (const char method, const char msgfmt,...) const
	Issue info message. More...

virtual void	Warning (const char method, const char msgfmt,...) const
	Issue warning message. More...

virtual void	Error (const char method, const char msgfmt,...) const
	Issue error message. More...

virtual void	SysError (const char method, const char msgfmt,...) const
	Issue system error message. More...

virtual void	Fatal (const char method, const char msgfmt,...) const
	Issue fatal error message. More...

void	AbstractMethod (const char *method) const
	Use this method to implement an "abstract" method that you don't want to leave purely abstract. More...

void	MayNotUse (const char *method) const
	Use this method to signal that a method (defined in a base class) may not be called in a derived class (in principle against good design since a child class should not provide less functionality than its parent, however, sometimes it is necessary). More...

void	Obsolete (const char method, const char asOfVers, const char *removedFromVers) const
	Use this method to declare a method obsolete. More...

Private Member Functions
	TPacketizerAdaptive ()

	TPacketizerAdaptive (const TPacketizerAdaptive &)

void	InitStats ()
	(re)initialise the statistics called at the begining or after a worker dies. More...

void	operator= (const TPacketizerAdaptive &)

TFileNode *	NextNode ()
	Get next node which has unallocated files. More...

void	RemoveUnAllocNode (TFileNode *)
	Remove unallocated node. More...

TFileNode *	NextActiveNode ()
	Get next active node. More...

void	RemoveActiveNode (TFileNode *)
	Remove node from the list of actives. More...

TFileStat *	GetNextUnAlloc (TFileNode node=0, const char nodeHostName=0)
	Get next unallocated file from 'node' or other nodes: First try 'node'. More...

TFileStat *	GetNextActive ()
	Get next active file. More...

void	RemoveActive (TFileStat *file)
	Remove file from the list of actives. More...

void	Reset ()
	Reset the internal data structure for packet distribution. More...

void	ValidateFiles (TDSet dset, TList slaves, Long64_t maxent=-1, Bool_t byfile=kFALSE)
	Check existence of file/dir/tree an get number of entries. More...

Int_t	ReassignPacket (TDSetElement e, TList *listOfMissingFiles)
	The file in the listOfMissingFiles can appear several times; in order to fix that, a TDSetElement::Merge method is needed. More...

void	SplitPerHost (TList elements, TList *listOfMissingFiles)
	Split into per host entries The files in the listOfMissingFiles can appear several times; in order to fix that, a TDSetElement::Merge method is needed. More...

Additional Inherited Members
Public Types inherited from TVirtualPacketizer
enum	EStatusBits { kIsInitializing = BIT(16), kIsDone = BIT(17), kIsTree = BIT(18) }

Public Types inherited from TObject
enum	EStatusBits { kCanDelete = BIT(0), kMustCleanup = BIT(3), kObjInCanvas = BIT(3), kIsReferenced = BIT(4), kHasUUID = BIT(5), kCannotPick = BIT(6), kNoContextMenu = BIT(8), kInvalidObject = BIT(13) }

enum	{ kIsOnHeap = 0x01000000, kNotDeleted = 0x02000000, kZombie = 0x04000000, kBitMask = 0x00ffffff }

enum	{ kSingleKey = BIT(0), kOverwrite = BIT(1), kWriteDelete = BIT(2) }

Static Public Member Functions inherited from TObject
static Long_t	GetDtorOnly ()
	Return destructor only flag. More...

static void	SetDtorOnly (void *obj)
	Set destructor only flag. More...

static Bool_t	GetObjectStat ()
	Get status of object stat flag. More...

static void	SetObjectStat (Bool_t stat)
	Turn on/off tracking of objects in the TObjectTable. More...

Protected Types inherited from TVirtualPacketizer
enum	EUseEstOpt { kEstOff = 0, kEstCurrent = 1, kEstAverage = 2 }

Protected Member Functions inherited from TVirtualPacketizer
	TVirtualPacketizer (TList input, TProofProgressStatus st=0)

	TVirtualPacketizer (const TVirtualPacketizer &)

void	operator= (const TVirtualPacketizer &)

TDSetElement *	CreateNewPacket (TDSetElement *base, Long64_t first, Long64_t num)
	Creates a new TDSetElement from from base packet starting from the first entry with num entries. More...

Long64_t	GetEntries (Bool_t tree, TDSetElement *e)
	Get entries. More...

virtual Bool_t	HandleTimer (TTimer *timer)
	Send progress message to client. More...

Protected Member Functions inherited from TObject
void	MakeZombie ()

virtual void	DoError (int level, const char location, const char fmt, va_list va) const
	Interface to ErrorHandler (protected). More...

Protected Attributes inherited from TVirtualPacketizer
Double_t	fMinPacketTime

Double_t	fMaxPacketTime

TList *	fConfigParams

TMap *	fSlaveStats

TProofProgressStatus *	fProgressStatus

TTimer *	fProgress

Long64_t	fTotalEntries

TList *	fFailedPackets

TTime	fStartTime

Float_t	fInitTime

Float_t	fProcTime

Float_t	fTimeUpdt

TNtupleD *	fCircProg

Long_t	fCircN

TNtuple *	fProgressPerf

Float_t	fProcTimeLast

Int_t	fActWrksLast

Float_t	fEvtRateLast

Float_t	fMBsReadLast

Float_t	fEffSessLast

Bool_t	fAWLastFill

Float_t	fReportPeriod

EUseEstOpt	fUseEstOpt

Bool_t	fValid

Bool_t	fStop

TString	fDataSet

TList *	fInput

#include <TPacketizerAdaptive.h>

Inheritance diagram for TPacketizerAdaptive:

[legend]

Constructor & Destructor Documentation

TPacketizerAdaptive::TPacketizerAdaptive ( )

private

TPacketizerAdaptive::TPacketizerAdaptive ( const TPacketizerAdaptive & )

private

TPacketizerAdaptive::TPacketizerAdaptive	(	TDSet *	dset,
		TList *	slaves,
		Long64_t	first,
		Long64_t	num,
		TList *	input,
		TProofProgressStatus *	st
	)

TPacketizerAdaptive::~TPacketizerAdaptive ( )

virtual

Destructor.

Definition at line 880 of file TPacketizerAdaptive.cxx.

Member Function Documentation

Int_t TPacketizerAdaptive::AddProcessed	(	TSlave *	sl,
		TProofProgressStatus *	status,
		Double_t	latency,
		TList **	listOfMissingFiles = `0`
	)

virtual

To be used by GetNextPacket but also in reaction to kPROOF_STOPPROCESS message (when the worker was asked to stop processing during a packet).

returns the #entries intended in the last packet - #processed entries

Reimplemented from TVirtualPacketizer.

Definition at line 1569 of file TPacketizerAdaptive.cxx.

Int_t TPacketizerAdaptive::CalculatePacketSize	(	TObject *	slstat,
		Long64_t	cachesz,
		Int_t	learnent
	)

The result depends on the fStrategy.

Definition at line 1479 of file TPacketizerAdaptive.cxx.

Referenced by GetNextPacket().

Int_t TPacketizerAdaptive::GetActiveWorkers ( )

virtual

Return the number of workers still processing.

Reimplemented from TVirtualPacketizer.

Definition at line 1949 of file TPacketizerAdaptive.cxx.

Float_t TPacketizerAdaptive::GetCurrentRate ( Bool_t & all )

virtual

Get Estimation of the current rate; just summing the current rates of the active workers.

Reimplemented from TVirtualPacketizer.

Definition at line 1966 of file TPacketizerAdaptive.cxx.

Int_t TPacketizerAdaptive::GetEstEntriesProcessed	(	Float_t	t,
		Long64_t &	ent,
		Long64_t &	bytes,
		Long64_t &	calls
	)

virtual

Get estimation for the number of processed entries and bytes read at time t, based on the numbers already processed and the latests worker measured speeds.

If t <= 0 the current time is used. Only the estimation for the entries is currently implemented. This is needed to smooth the instantaneous rate plot.

Reimplemented from TVirtualPacketizer.

Definition at line 1995 of file TPacketizerAdaptive.cxx.

TPacketizerAdaptive::TFileStat * TPacketizerAdaptive::GetNextActive ( )

private

Get next active file.

Definition at line 1047 of file TPacketizerAdaptive.cxx.

TDSetElement * TPacketizerAdaptive::GetNextPacket	(	TSlave *	sl,
		TMessage *	r
	)

virtual

Get next packet; A meaningfull difference to TPacketizer is the fact that this packetizer, for each worker, tries to predict whether the worker will finish processing it's local files before the end of the query.

If yes, it allocates, to those workers, files from non-slave filenodes or from slaves that are overloaded. The check is done every time a new file needs to be assigned.

Reimplemented from TVirtualPacketizer.

Definition at line 1666 of file TPacketizerAdaptive.cxx.

TPacketizerAdaptive::TFileStat * TPacketizerAdaptive::GetNextUnAlloc	(	TFileNode *	node = `0`,
		const char *	nodeHostName = `0`
	)

private

Get next unallocated file from 'node' or other nodes: First try 'node'.

If there is no more files, keep trying to find an unallocated file on other nodes.

Definition at line 935 of file TPacketizerAdaptive.cxx.

void TPacketizerAdaptive::InitStats ( )

private

(re)initialise the statistics called at the begining or after a worker dies.

Definition at line 897 of file TPacketizerAdaptive.cxx.

Referenced by ClassImp(), and MarkBad().

void TPacketizerAdaptive::MarkBad	(	TSlave *	s,
		TProofProgressStatus *	status,
		TList **	listOfMissingFiles
	)

virtual

This method can be called at any time during processing as an effect of handling kPROOF_STOPPROCESS If the output list from this worker is going to be sent back to the master, the 'status' includes the number of entries processed by the slave.

From this we calculate the remaining part of the packet. 0 indicates that the results from that worker were lost completely. Assume that the filenodes for which we have a TFileNode object are still up and running.

Reimplemented from TVirtualPacketizer.

Definition at line 2070 of file TPacketizerAdaptive.cxx.

TPacketizerAdaptive::TFileNode * TPacketizerAdaptive::NextActiveNode ( )

private

Get next active node.

Definition at line 1064 of file TPacketizerAdaptive.cxx.

TPacketizerAdaptive::TFileNode * TPacketizerAdaptive::NextNode ( )

private

Get next node which has unallocated files.

the order is determined by TFileNode::Compare

Definition at line 1018 of file TPacketizerAdaptive.cxx.

Referenced by GetNextUnAlloc().

void TPacketizerAdaptive::operator= ( const TPacketizerAdaptive & )

private

Int_t TPacketizerAdaptive::ReassignPacket	(	TDSetElement *	e,
		TList **	listOfMissingFiles
	)

private

The file in the listOfMissingFiles can appear several times; in order to fix that, a TDSetElement::Merge method is needed.

Definition at line 2130 of file TPacketizerAdaptive.cxx.

Referenced by AddProcessed(), and SplitPerHost().

void TPacketizerAdaptive::RemoveActive ( TFileStat * file )

private

Remove file from the list of actives.

Definition at line 1086 of file TPacketizerAdaptive.cxx.

void TPacketizerAdaptive::RemoveActiveNode ( TFileNode * node )

private

Remove node from the list of actives.

Definition at line 1097 of file TPacketizerAdaptive.cxx.

void TPacketizerAdaptive::RemoveUnAllocNode ( TFileNode * node )

private

Remove unallocated node.

Definition at line 1039 of file TPacketizerAdaptive.cxx.

void TPacketizerAdaptive::Reset ( void )

private

Reset the internal data structure for packet distribution.

Definition at line 1105 of file TPacketizerAdaptive.cxx.

void TPacketizerAdaptive::SplitPerHost	(	TList *	elements,
		TList **	listOfMissingFiles
	)

private

Split into per host entries The files in the listOfMissingFiles can appear several times; in order to fix that, a TDSetElement::Merge method is needed.

Definition at line 2175 of file TPacketizerAdaptive.cxx.

Referenced by MarkBad().

void TPacketizerAdaptive::ValidateFiles	(	TDSet *	dset,
		TList *	slaves,
		Long64_t	maxent = `-1`,
		Bool_t	byfile = `kFALSE`
	)

private

Check existence of file/dir/tree an get number of entries.

Assumes the files have been setup.

Definition at line 1154 of file TPacketizerAdaptive.cxx.

Member Data Documentation

TList* TPacketizerAdaptive::fActive

private

Definition at line 60 of file TPacketizerAdaptive.h.

Float_t TPacketizerAdaptive::fBaseLocalPreference

private

Definition at line 72 of file TPacketizerAdaptive.h.

Referenced by ClassImp(), and GetNextPacket().

Bool_t TPacketizerAdaptive::fCachePacketSync

private

Definition at line 66 of file TPacketizerAdaptive.h.

Referenced by CalculatePacketSize(), and ClassImp().

TList* TPacketizerAdaptive::fFileNodes

private

Definition at line 55 of file TPacketizerAdaptive.h.

TSortedList* TPacketizerAdaptive::fFilesToProcess

private

Definition at line 64 of file TPacketizerAdaptive.h.

Referenced by CalculatePacketSize(), ClassImp(), GetNextPacket(), and ~TPacketizerAdaptive().

Bool_t TPacketizerAdaptive::fForceLocal

private

Definition at line 74 of file TPacketizerAdaptive.h.

Referenced by ClassImp(), and GetNextPacket().

Float_t TPacketizerAdaptive::fFractionOfRemoteFiles

private

Definition at line 69 of file TPacketizerAdaptive.h.

Referenced by InitStats().

Double_t TPacketizerAdaptive::fMaxEntriesRatio

private

Definition at line 67 of file TPacketizerAdaptive.h.

Referenced by CalculatePacketSize(), and ClassImp().

Int_t TPacketizerAdaptive::fMaxPerfIdx

private

Definition at line 61 of file TPacketizerAdaptive.h.

Long_t TPacketizerAdaptive::fMaxSlaveCnt

private

Definition at line 76 of file TPacketizerAdaptive.h.

Long64_t TPacketizerAdaptive::fNEventsOnRemLoc

private

Definition at line 70 of file TPacketizerAdaptive.h.

Referenced by GetNextPacket(), and InitStats().

Int_t TPacketizerAdaptive::fPacketAsAFraction

private

Definition at line 78 of file TPacketizerAdaptive.h.

TList* TPacketizerAdaptive::fPartitions

private

Definition at line 62 of file TPacketizerAdaptive.h.

Int_t TPacketizerAdaptive::fStrategy

private

Definition at line 84 of file TPacketizerAdaptive.h.

Referenced by CalculatePacketSize(), ClassImp(), and GetNextPacket().

Int_t TPacketizerAdaptive::fTryReassign

private

Definition at line 85 of file TPacketizerAdaptive.h.

Referenced by ClassImp(), and ReassignPacket().

TList* TPacketizerAdaptive::fUnAllocated

private

Definition at line 59 of file TPacketizerAdaptive.h.

Collaboration diagram for TPacketizerAdaptive:

[legend]

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

proof/proofplayer/inc/TPacketizerAdaptive.h
proof/proofplayer/src/TPacketizerAdaptive.cxx

Private Attributes
TList *	fFileNodes

TList *	fUnAllocated

TList *	fActive

Int_t	fMaxPerfIdx

TList *	fPartitions

TSortedList *	fFilesToProcess

Bool_t	fCachePacketSync

Double_t	fMaxEntriesRatio

Float_t	fFractionOfRemoteFiles

Long64_t	fNEventsOnRemLoc

Float_t	fBaseLocalPreference

Bool_t	fForceLocal

Long_t	fMaxSlaveCnt

Int_t	fPacketAsAFraction

Int_t	fStrategy

Int_t	fTryReassign

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation