Unverified Commit 60495b40 authored by Eisenhauer, Greg's avatar Eisenhauer, Greg Committed by GitHub
Browse files

Merge pull request #999 from eisenhauer/SstCosmetic 

Docs, verbosity and code clarity changes
parents 8e11d71d 5ed75551

docs/user_guide/source/engines/sst.rst

+29 −12
Original line number Diff line number Diff line
@@ -136,18 +136,34 @@ satisfying, but has a similar long-term effect upon the set of steps 
delivered to the readers.)  This value is interpreted by SST Writer engines

only.



5. **DataTransport**:  Default **"RDMA"**.   This string value specifies the

underlying network communication mechanism to use for exchanging data in

SST.  Current allowed values are **"RDMA"** and **"WAN"**.  (**ib** and

**fabric** are accepted as equivalent to **RDMA** and **evpath** is

equivalent to **WAN**.)  Generally both the reader and writer should be

using the same network transport, and the network transport chosen may be

dictated by the situation.  For example, the RDMA transport generally

operates only between applications running on the same high-performance

interconnect (e.g. on the same HPC machine).  If communication is desired

between applications running on different interconnects, the Wide Area

Network (WAN) option should be chosen.  This value is interpreted by both

SST Writer and Reader engines.

5. **DataTransport**: Default **varies**.  This string value specifies

the underlying network communication mechanism to use for exchanging

data in SST.  Generally this is chosen by SST based upon what is

available on the current platform.  However, specifying this engine

parameter allows overriding SST's choice.  Current allowed values are

**"RDMA"** and **"WAN"**.  (**ib** and **fabric** are accepted as

equivalent to **RDMA** and **evpath** is equivalent to **WAN**.)

Generally both the reader and writer should be using the same network

transport, and the network transport chosen may be dictated by the

situation.  For example, the RDMA transport generally operates only

between applications running on the same high-performance interconnect

(e.g. on the same HPC machine).  If communication is desired between

applications running on different interconnects, the Wide Area Network

(WAN) option should be chosen.  This value is interpreted by both SST

Writer and Reader engines.



5. **NetworkInterface**: Default **NULL**.  In situations in which

there are multiple possible network interfaces available to SST, this

string value specifies which should be used to generate SST's contact

information for writers.  Generally this should *NOT* be specified

except for narrow sets of circumstances.  It has no effect if

specified on Reader engines.  If specified, the string value should

correspond to a name of a network interface, such as are listed by

commands like "netstat -i".  For example, on most Unix systems,

setting the NetworkInterface parameter to "lo" (or possibly "lo0")

will result in SST generating contact information that uses the

network address associated with the loopback interface (127.0.0.1).

This value is interpreted by only by the SST Writer engine.



====================   ===================== =========================================================

 **Key**                **Value Format**      **Default** and Examples
@@ -157,4 +173,5 @@ SST Writer and Reader engines. 
 QueueLimit            integer		     **0** (no queue limits)

 QueueFullPolicy       string	             **Block**, Discard

 DataTransport	       string		     **default varies by platform**, RDMA, WAN

 NetworkInterface      string		     **NULL**

====================   ===================== =========================================================

source/adios2/engine/sst/SstReader.cpp

+18 −2
Original line number Diff line number Diff line
@@ -116,13 +116,29 @@ SstReader::SstReader(IO &io, const std::string &name, const Mode mode, 


SstReader::~SstReader() { SstStreamDestroy(m_Input); }



StepStatus SstReader::BeginStep(StepMode mode, const float timeout_sec)

StepStatus SstReader::BeginStep(StepMode Mode, const float timeout_sec)

{



    SstStatusValue result;

    SstStepMode StepMode;

    switch (Mode)

    {

    case adios2::StepMode::Append:

    case adios2::StepMode::Update:

        throw std::invalid_argument(

            "ERROR: SstReader::BeginStep inappropriate StepMode specified" +

            m_EndMessage);

        break;

    case adios2::StepMode::NextAvailable:

        StepMode = SstNextAvailable;

        break;

    case adios2::StepMode::LatestAvailable:

        StepMode = SstLatestAvailable;

        break;

    }

    m_IO.RemoveAllVariables();

    m_IO.RemoveAllAttributes();

    result = SstAdvanceStep(m_Input, (int)mode, timeout_sec);

    result = SstAdvanceStep(m_Input, StepMode, timeout_sec);

    if (result == SstEndOfStream)

    {

        return StepStatus::EndOfStream;

source/adios2/toolkit/sst/cp/cp_internal.h

+3 −0
Original line number Diff line number Diff line
@@ -50,6 +50,9 @@ enum StreamStatus 
    Closed

};



static char *SSTStreamStatusStr[] = {"NotOpen", "Established", "PeerClosed",

                                     "PeerFailed", "Closed"};



typedef struct _WS_ReaderInfo

{

    SstStream ParentStream;

source/adios2/toolkit/sst/cp/cp_reader.c

+8 −7
Original line number Diff line number Diff line
@@ -153,8 +153,8 @@ static void ReaderConnCloseHandler(CManager cm, CMConnection ClosedConn, 
        fprintf(stderr, "Got an unexpected connection close event\n");

        CP_verbose(Stream, "Reader-side Rank received a "

                           "connection-close event in unexpected "

                           "state %d\n",

                   Stream->Status);

                           "status %s\n",

                   SSTStreamStatusStr[Stream->Status]);

    }

}
@@ -762,7 +762,8 @@ static TSMetadataList waitForNextMetadata(SstStream Stream, long LastTimestep) 
        CP_verbose(Stream,

                   "Waiting for metadata for a Timestep later than TS %d\n",

                   LastTimestep);

        CP_verbose(Stream, "Stream status is %d\n", Stream->Status);

        CP_verbose(Stream, "(PID %x) Stream status is %s\n", getpid(),

                   SSTStreamStatusStr[Stream->Status]);

        /* wait until we get the timestep metadata or something else changes */

        pthread_cond_wait(&Stream->DataCondition, &Stream->DataLock);

    }
@@ -880,7 +881,7 @@ extern void SstReleaseStep(SstStream Stream) 
 * wait for metadata for Timestep indicated to arrive, or fail with EndOfStream

 * or Error

 */

extern SstStatusValue SstAdvanceStep(SstStream Stream, int mode,

extern SstStatusValue SstAdvanceStep(SstStream Stream, SstStepMode mode,

                                     const float timeout_sec)

{
@@ -892,7 +893,7 @@ extern SstStatusValue SstAdvanceStep(SstStream Stream, int mode, 
        Stream->CurrentMetadata = NULL;

    }



    if ((timeout_sec >= 0.0) || (mode == 3 /* LatestAvailable*/))

    if ((timeout_sec >= 0.0) || (mode == SstLatestAvailable))

    {

        struct _GlobalOpInfo

        {
@@ -973,7 +974,7 @@ extern SstStatusValue SstAdvanceStep(SstStream Stream, int mode, 
                 * others will see shortly.  I'm going to go with Biggest

                 * until I have a reason to prefer one or the other.

                */

                if (mode == 3)

                if (mode == SstLatestAvailable)

                {

                    // latest available

                    CP_verbose(Stream, "Returning Biggest timestep available "
@@ -1009,7 +1010,7 @@ extern SstStatusValue SstAdvanceStep(SstStream Stream, int mode, 
            CP_verbose(Stream, "Advancestep timing out on no data\n");

            return SstTimeout;

        }

        if (mode == 3)

        if (mode == SstLatestAvailable)

        {

            // latest available

            /* release all timesteps from before NextTimestep, then fall

source/adios2/toolkit/sst/cp/cp_writer.c

+6 −4
Original line number Diff line number Diff line
@@ -230,8 +230,8 @@ static void WriterConnCloseHandler(CManager cm, CMConnection closed_conn, 
        fprintf(stderr, "Got an unexpected connection close event\n");

        CP_verbose(ParentWriterStream, "Writer-side Rank received a "

                                       "connection-close event in unexpected "

                                       "state %d\n",

                   WSreader->ReaderStatus);

                                       "state %s\n",

                   SSTStreamStatusStr[WSreader->ReaderStatus]);

        PTHREAD_MUTEX_UNLOCK(&ParentWriterStream->DataLock);

        CP_PeerFailCloseWSReader(WSreader, PeerFailed);

        PTHREAD_MUTEX_LOCK(&ParentWriterStream->DataLock);
@@ -833,6 +833,8 @@ static void CP_PeerFailCloseWSReader(WS_ReaderInfo CP_WSR_Stream, 
        CMadd_delayed_task(ParentStream->CPInfo->cm, 2, 0, CloseWSRStream,

                           CP_WSR_Stream);

    }

    CP_verbose(ParentStream, "Moving stream %p to status %s\n", CP_WSR_Stream,

               SSTStreamStatusStr[NewState]);

    CP_WSR_Stream->ReaderStatus = NewState;

    /* main thread might be waiting on timesteps going away */

    pthread_cond_signal(&ParentStream->DataCondition);
@@ -876,8 +878,8 @@ void SstWriterClose(SstStream Stream) 
        CP_verbose(Stream, "Reader Count is %d\n", Stream->ReaderCount);

        for (int i = 0; i < Stream->ReaderCount; i++)

        {

            CP_verbose(Stream, "Reader [%d] status is %d\n", i,

                       Stream->Readers[i]->ReaderStatus);

            CP_verbose(Stream, "Reader [%d] status is %s\n", i,

                       SSTStreamStatusStr[Stream->Readers[i]->ReaderStatus]);

        }

        /* NEED TO HANDLE FAILURE HERE */

        pthread_cond_wait(&Stream->DataCondition, &Stream->DataLock);