airavata-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Saminda Wijeratne <>
Subject Exposing Workflow Context changes to the API user
Date Fri, 21 Dec 2012 23:11:55 GMT
This is a more descriptive explanation of the offline suggestions we got
for WorkflowContextHeader.

It was suggested that we should have a much simpler way to allow users to
define the Workflow Context information when running an Experiment via the
API. Currently what we have is the WorkflowContextHeaderBuilder [1] class
which contains the following context information for the user to change,
*Security Context* - customized security measures
*Service Endpoints* - use the user give service endpoints for interpreter
service/gfac/message box etc.
*Output data handling* - how to handle different intermediate output data
*Workflow Scheduling* - customized job scheduling options
*Monitoring Context* - <something to do with monitoring running workflow
Each of the above are xml schema objects which users are required to
manipulate to define the custom workflow context settings.

In order to simplify following was recommended to be exposed as separate
Workflow Scheduling
Output data handling
Security Context

I started creating beans for Workflow Scheduling as following

WorkflowSchedulingSettings [2]
|-node Id
|-service Id
|-host settings
|    |-host id
|    |-use ws-gram
|    |_gatekeeper epr
|_hpc settings
|-job manager
|-cpu count
|-node count
|-queue name
|_max wall time

I've added relevant getters/setters/adders/re-setters to manage above data
for workflow scheduling. Your thoughts are welcome for improvements.



  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message