Solving a Mosel model using the 'executor' package, from Mosel
In this example, the Xpress Executor should first be configured with the blend3c.mos model. Then, you run the blendexecutor.mos model locally; this uses the 'executor' package to send the blend.csv file to the Xpress Executor and remotely solve the model using this, then downloads and displays the results. This example requires a local installation of Xpress.
(!****************************************************** Xpress Executor Example Model ============================= file blendexecutor.mos `````````````````````` Demonstrates executing the 'blend3c' model using Xpress Executor with the supplied input data & displaying the results. Uses the 'executor' package included in Xpress to talk to Xpress Executor. Assumes that you have an Xpress Executor component already configured with the blend3c model. (c) 2017-2017 Fair Isaac Corporation author: J. Farmer, Oct. 2017 *******************************************************!) model "Blend Executor" uses "mmsystem", "mmssl", "executor" parameters ! You should set the DMP_EXECUTOR_URL, DMP_SOLUTION_CLIENT_ID and DMP_SOLUTION_SECRET parameters ! to point to the component you want to test. ! The root URL of the Xpress Executor DMP component ! Obtain this by clicking "View Links" for the Xpress Executor component on the DMP UI, then remove the ! part of the URL after the domain name, e.g.: https://vm65j75lqh-vm65j75lqh.dms.usw2.ficoanalyticcloud.com DMP_EXECUTOR_URL="" ! The client ID of solution containing the Xpress Executor DMP component ! Obtain this through the DMP UI DMP_SOLUTION_CLIENT_ID="" ! The secret of the solution containing the Xpress Executor DMP component ! Obtain this through the DMP UI DMP_SOLUTION_SECRET="" ! The root DMP authentication URL. This will be different depending on which instance of DMP you are using. DMP_BEARER_TOKEN_ROOT_URL="https://iam-svc.dms.usw2.ficoanalyticcloud.com/" ! The input file for the remote model INPUTFILE="../data/blend.csv" ! File into which to save the results RESULTFILE="blendresults.csv" end-parameters declarations ! Entity representing our Xpress Executor component instance myExecutor: Executor ! Record representing our remote execution and its current status modelExecution: ModelExecution ! Array of parameters modelParameterNames: set of string modelParameters: dynamic array(modelParameterNames) of text end-declarations public declarations public inputText: text public runLog: text end-declarations ! Declare array used to hold result data declarations ORES=1..2 use: array(ORES) of real end-declarations ! Configure Mosel to use TLS1.2 ciphers with HTTPS to allow us to talk to DMP setparam("https_ciphers","TLSv1.2+HIGH:\!SSLv2:\!aNULL:\!eNULL:\!3DES:@STRENGTH") ! Log into DMP writeln("Initializing executor entity") myExecutor.componenturl := DMP_EXECUTOR_URL myExecutor.dmpmanagerurl := DMP_BEARER_TOKEN_ROOT_URL ! Note despite the field name, this should be populated with the root of the URL for obtaining the bearer token, ! as described in the DMP documentation, which can be different from the DMP Manager URL myExecutor.clientid := DMP_SOLUTION_CLIENT_ID myExecutor.secret := DMP_SOLUTION_SECRET ! Execute our model writeln("Submitting model execution") modelParameters("INPUTFILE"):="input" ! In Xpress Executor, the inputs we provide will be provided to the model in a file called "input" modelParameters("RESULTFILE"):="result" ! In Xpress Executor, we can only access model results from a file called "result" modelExecution := executorexecute( myExecutor, INPUTFILE, modelParameters ) if myExecutor.status<>EXECUTOR_OK then writeln("Error starting execution: ",myExecutor.lasterror) exit(1) end-if ! Model will be executing asynchronously; wait up to 10 minutes for it to complete. writeln("Waiting for completion of execution") executorwaitfor( myExecutor, modelExecution, 10*60 ) if myExecutor.status<>EXECUTOR_OK then writeln("Error waiting for execution: ",myExecutor.lasterror) ! myExecutor.status OK means we had no errors querying the status and either the model has completed or ! the model is still running after our timeout elif not modelExecution.iscompleted then writeln("Execution failed to complete after 10 minutes") ! Check if the model status isn't OK or the exit code isn't zero ! Depending on what your model does, you may not need to check the exit code or a non-zero value may not indicate ! an error, but an status other than OK always means an error. elif modelExecution.status<>EXECUTION_STATUS_OK or modelExecution.exitcode<>0 then writeln("Execution failed, status=",modelExecution.status," and exit code=", modelExecution.exitcode) ! In event of failure, output the run log so we can diagnose the problem executorfetchrunlog( myExecutor, modelExecution, "text:runLog" ) if myExecutor.status<>EXECUTOR_OK then writeln("Error fetching run log: ",myExecutor.lasterror) else writeln("Execution run log follows:") writeln(runLog) end-if ! If model execution status is OK and exitCode is 0, execution was successful so download and display results else writeln("Fetching execution results") executorfetchresult( myExecutor, modelExecution, RESULTFILE ) if myExecutor.status<>EXECUTOR_OK then writeln("Error fetching result file: ",myExecutor.lasterror) else ! Parse results and display to user initializations from "mmsheet.csv:"+RESULTFILE use as "[A1:B2]" end-initializations writeln writeln("Solution:") forall(o in ORES) writeln(" use(",o,"): ",use(o)) end-if end-if ! Whether we were successful or not, delete execution from component executordelete( myExecutor, modelExecution ) if myExecutor.status<>EXECUTOR_OK then writeln("Failed to delete execution due to error: ", myExecutor.lasterror) exit(1) end-if end-model
|© Copyright 2021 Fair Isaac Corporation.|