Time:
1:00 PM Eastern US (contact
MichaelFiedler if you'd like to participate)
The Automation meetings alternate times each meeting to accomodate the global team.
Agenda
* Reoccurring agenda items:
* Main agenda items:
- Previous Action Items:
- Automation Contribution and Automation Result drill down discussion
- CharlesRankin, MaxVohlken, PaulMcMahan, PramodChandoria, VaibhavSrivastava and MichaelFiedler met on 5 December to discuss
- Discussion covered
- Modelling of the relationship. Do consumers need to be able to find contributions as independent artifacts? Sentiment was this was not likely. Always need them in context of a result.
- Need to allow for contributions from non-local automation providers.
- Need to allow for contributions in-line with an Automation Result and contributions which are referenced (via URL) from within the result
- Follow on action item to propose exactly what the artifacts will look like and see if they hold up to our scenarios.
- Main topic: Discuss the DevOps and Deployment scenarios
- Plans for moving from scenario development to spec development
- Next meetings:
Minutes
Attending: Michael Fiedler, Bill Higgins, Dan Berg, Max Vohlken, Rich Rakich , John Arwe, Srimanth Gunturi, Pramod Chandoria, Eric Bordeau, Vaibhav Srivastava, Paul Mcmahan. I know I missed a few others, please add your names.
* Recap of Automation Contribution discussion
*
DevOps? and deployment scenarios
- General observations
- Automation "template" plans can be cloned and customized with more specific environment/input data prior to requesting execution. How "read only" are automation plans?
- DevOps? automation scenarios are chained together. Build->Deploy in sandbox->Test->Deploy in production. The logical concept of passing results from one build to another is important.
- There is a need to determine the "subtype" of an Automation provider. In an environment like DevOps? , times when only interested in build providers - other times just interested in test providers, etc. Need to filter at provider level.
- Scenario 1 - deployment execution
- Full scenario represents multiple discrete automation plan executions chained together by the provider. Availability of previous stages results to current stage is crtical.
- Deployment operations are triggered by events - started automatically from automation plans with environment information pre-configured for the most part.
- Need the ability to store arbitrary environment data in an automation plan and automation request
- Discussion on whether name/value was still an acceptable approach. General consensus was that it likely still held up, but need to see actual proposed artifacts to validate
- "Posting activity events" == contributions to an automation result
- Comments that it would help to re-align the language in this scenario with the OSLC artifacts discussed to date
- Scenario 2 - deployment configuration
- When configuring an automation, need to be able to filter on certain automation provider types (build, test, deployment)
- "Configure the stage" == configuring the environment for the execution
- Policy aspects to this scenario are out of scope for OSLC spec
- however decision points require automation results be available to support policy decisions
* Concluding discussion revisiting the environment configuration
- Keeping the env configuration as simple as possible
- Need further discussion on "input" configuration vs "output" configuration in the results. Topic for next meeting
* Next meeting
- 15 December - final meeting for 2011
- Continue environment configuration discussion
- Possible conversation on troubleshooting sub-scenario (task drill-down)
- Will try to have some artifact proposals for initial review