110 likes | 234 Views
ALMA Integrated Computing Team Coordination & Planning Meeting #1 Santiago, 17-19 April 2013. Acceptance Manager Transition Steve Scott. Overview. Current AM term through end of 2013 Propose changes in process and duties Start transition soon With next acceptance – May/June
E N D
ALMA Integrated Computing TeamCoordination & Planning Meeting #1Santiago, 17-19 April 2013 Acceptance Manager Transition Steve Scott
Overview • Current AM term through end of 2013 • Propose changes in process and duties • Start transition soon • With next acceptance – May/June • Basic Principles • General flow of acceptance is unchanged, but… • Trim acceptance process as much as possible • Use JIRA as much as possible • Cut out the middle-man • Direct interaction of stakeholders (ICT/SciOps) • More responsibility for stakeholders (but not work) • More input (and responsibilities) from QM • Less work (and responsibilities) for AM (<= 25% FTE)
Communication • Twiki is focal point • ‘Everyone’ informed at start of acceptance process • Sign up for notifications if they want • Key stakeholders (‘committee’) are special • Meetings scheduled around their availability • But everyone welcome to attend! • Keep as small as possible (ease of scheduling) • AM • Computing (1 or 2?) • Science (1 or 2?) • ASG/AIV (1 rep only) for Online acceptances only
Communication 2 • Masao is the coordinator of the Offline Phase V testing/testers • Single point of contact for the AM • This will significantly reduce AM work load • New change, just now going into effect
Document Content • Feature list • Testing list and summary, links to reports • Documentation links • Links to acceptance tests • QM provides • Test failures • Bugs • Evaluation of quality • Conditions on previous acceptances • Maybe this is a separate doc, maybe not • Reduces AM work load
Document Changes Overview • Acceptance Plan (for TRR) • Only signed by AM • EDM • Acceptance Report • Signed by AM, Computing, Science (as now) • EDM • No Acceptance Certificate • Signed Acceptance Report is sufficient • All other docs (test reports…) on twiki (as now) • EDM would be nice, but…
AM as Broker • ICT • “I’ve got some great software, barely been used…” • “Look at all these fantastic test results!” • DSO • “I’ll test it by driving off a cliff, if it survives then I’ll take it” • QM • “The Odometer has been rolled back” • “The left rear tire is flat” • AM • “ICT, can you fix that tire and maybe the fender too?” • “DSO, maybe just drive it down to the beach?”
Producer/Consumer • Cut out the middle man • Basic responsibilities fall on producer and consumer • ICT/DSO negotiate start of acceptance • TRR • ICT makes case for their software • DSO states what tests will validate the software • Acceptance Review • DSO decision on acceptance, based on test results • ICT agrees with decision or makes it own case • AM provides process, bookkeeping
What to Keep • The basic acceptance process, with some minimal formalism, gives order and known expectations • Open process, everyone welcome • The principle that software should be accepted before it is deployed for official observatory use • Data schema changes should not force deployment of new software • SCCB provides stability
Process Improvements • JIRA field in feature for acceptance version, e.g. ‘A9.1R4’ • Define acceptance tests early (before Phase I) • Developers will ensure that code passes • Acceptance testing is the integrated test • Phase V testing should be feature oriented • Pass/fail criteria clearly stated • For Online, a 1 or 2 night pre-TRR test to make sure we are ready for TRR (until regression tests are complete), with a week of acceptance testing after TRR
Transition Schedule • Replace AM on SCCB with Science rep • Cut out the middle man • It’s not much work • Make change by the end of this meeting • Start with lean acceptance process in May/June • Bring on next AM in acceptance after the May/June • Will work with current AM through end of year, ramping up and sharing duties • New definition of AM duties may open up position to more candidates