Module percept2

Description
Data Types
Function Index
Function Details

Percept2 - An Enhanced version of the Erlang Concurrency Profiling Tool Percept.

Behaviours: application.

Description

Percept2 - An Enhanced version of the Erlang Concurrency Profiling Tool Percept.

Percept2 extends Percept in two aspects: functionality and scalability. Among the new functionalities added to Percept are:

Scheduler activity: the number of active schedulers at any time.
Process migration information: the migration history of a process between run queues.
Statistics data about message passing between processes: the number of messages, and the average message size, sent/received by a process.
Accumulated runtime per-process: the accumulated time when a process is in a running state.
Process tree: the hierarchy structure indicating the parent-child relationships between processes.
Dynamic function call graph/count/time: the hierarchy structure showing the calling relationships between functions during the program run, and the amount of time spent on a function.
Active functions: the functions that are active during a specific time interval.
Inter-node message passing: the sending of messages from one node to another.

The following techniques have been used to improved the scalability of Percept.

Compressed process tree/function call graph representation: an approach to reducing the number of processes/function call paths presented without losing important information.
Parallelisation of Percept: the processing of profile data has been parallelised so that multiple data files can be processed at the same time

This module provides the user interface for the application.

Data Types

filespec()

filespec() = file:filename() | {file:filename(), wrap, Suffix::string(), WrapSize::pos_integer(), WrapCnt::pos_integer()}

trace_profile_option()

trace_profile_option() = procs | ports | schedulers | running | message | migration | all | {callgraph, [module_name()]}

Function Index

analyze/1	Parallel analysis of trace files.
analyze/4	Parallel analysis of trace files.
profile/2	Starts the profiling while an application of interest is already running.
profile/3	The profiling starts with executing the entry function given, and goes on for the whole duration until the entry function returns and the profiling has concluded.
start_webserver/0	Starts webserver.
start_webserver/1	Starts webserver with a given port number.
stop_db/0	Stops the percept2 database.
stop_profile/0	Stops the profiling.
stop_webserver/0	Stops webserver.
stop_webserver/1	Stops webserver of the given port.

Function Details

analyze/1

analyze(FileNames::[file:filename()]) -> ok | {error, any()}

Parallel analysis of trace files. See the Overview page for examples.

analyze/4

analyze(Filename::file:filename(), Suffix::string(), StartIndex::pos_integer(), EndIndex::pos_integer()) -> ok | {error, any()}

Parallel analysis of trace files. See the Overview page for examples.

profile/2

profile(FileSpec::filespec(), TraceProfileOptions::[trace_profile_option()]) -> {ok, integer()} | {already_started, port()}

Starts the profiling while an application of interest is already running. The profiling can be stopped by percept2:stop_profile/0. The valid TraceProfileOptions values are: procs, ports, schedulers, running, message, migration and all. See profile/3 for the descriptions of the options.

See also: stop_profile/0.

profile/3

profile(FileSpec::filespec(), Entry::{module_name(), function_name(), [term()]}, TraceProfileOptions::[trace_profile_option()]) -> ok | {already_started, port()}

The profiling starts with executing the entry function given, and goes on for the whole duration until the entry function returns and the profiling has concluded. The events to be traced/profiled depends on the options specified by TraceProfileOptions. The following options are available:

-- procs : enables the profiling of process concurrency.

-- ports : enables the profiling of port concurrecny.

-- schedulers : enables the profiling of scheduler concurrency.

-- running : enables the feature to distinguish running from runnable process states. If the procs option is not given, this option enables the process concurrency automatically.

-- message : this enables the profiling of message passing between processes; If the procs option is not given, this option enables the process concurrency automatically.

-- migration : this enables the profiling of process migration between schedulers; If the procs option is not given, this option enables the process concurrency automatically.

-- all : this enable all the previous options.

-- {callgraph, Mods} : This enables the profiling of function activities (call and return_to) of functions defined in Mods. If the procs option is not given, this option enables the process concurrency automatically. Given the huge amount of data that could possibly be produced when this feature is on, we do not recommend profiling many modules in one go at this stage.

See the Overview page for examples.