Tuesday 3 December 2013

some output from few days



Through lot of shell scripts (linux, windows) and spring xmls, and after few experiments cause of issue was found:
on linux:

diff ../config/script/environment.properties.good ../config/script/environment_properties.bad 
65c65
---
  CAS_RS_CLASSPATH=/media/ephemeral0/endeca/CAS/3.1.2.1//lib/recordstore-forge-adapter/recordstore-forge-adapter-3.1.2.jar

on windows:

diff ../config/script/environment.properties.good ../config/script/environment_properties.bad 
65c65
  CAS_RS_CLASSPATH=/media/ephemeral0/endeca/CAS/3.1.2.1//lib/recordstore-forge-adapter/recordstore-forge-adapter-3.1.2.jar
---


Context of my work: automated deployment of enterprise apps { Endeca , ATG Web Commerce } by using proprietary PaaS engine ; instructions validation; knowledge transfer.

Issue was look like an  java.lang.ClassNotFoundException in process of Forge configuration {config}: (AdapterRunner): Error creating bean with name 'schemaRecordStoreAdapter' defined in class path
 resource [config/script/fcm-record-stores.context.xml]: Cannot create inner bean 'com.endeca.itl.recordstore.forge.RecordStoreSource
Bad environment.properties was generated by using CIM ( ATG web commerce) ,good was generated by  endeca tools and framework deployment_template deploy.sh.

It is a small one routine issue, next one is more interesting : - )

This part of my work was based on http://iam-saminda.blogspot.ru/2013/06/oracle-ecommerce-atg-102.html and oracle atg 10.2 documentations. Thanks for all who are working on such materials. 

Monday 4 November 2013

Software development, sometimes it is something like  data miming process or studying. Sometimes SD it is more like knowledge transfer.  

Saturday 2 November 2013

just a status update , just for myself

My current work/project is focused on deployment automation in context of software development  processes (CI CD processes). And there is no clear understanding what part of CI CD processes should be automated. There is only two main directions/ideas: 1) deployment automation it is baseline 2) deployment automation without integration with all development processes does not work and does not have value. Why ? Goal is to accelerate software development.  from my point of view such goal definition is very (or maybe even too ) ambitious .
One of the first results is opensourced Behavior Analytic Starter Kit ( recommendation engine on Hadoop Mahout, deployment automation of Hadoop cluster ( HDFS, MapReduce) + e-shop based on Broadleaf framework) https://github.com/griddynamics/Behavior-Analytic-Starter-Kit. Second one is private( at this moment) ATG deployment automation.