Skip to main content

Notice: this Wiki will be going read only early in 2024 and edits will no longer be possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.

Jump to: navigation, search

TPTP-PMC-20071212

Attending: Oliver, Guru, Chris, AlexN, AlexA, Paul, Joanna

Logistics

  • Minutes from last week showed up on Monday
    • Everyone should look at them
  • Guru needs to leave early
    • Asked to move Trace related items early

Profiler of the Gods (POG)

Alex reported on profiler initiative -- Started working with Harm (first lead user). Status:

  • Use case that he proposed has been defined
  • Alex is able to reproduce on host
    • Alex raised the question of whether we are trying to help folks be effective on their first use of tool or on their common use tasks.
      • Alex suggests separating these into two different areas and pursuing them independently.

Oliver asked whether M4 will have better experience than previous version.

  • Alex believes it will be better because of bug fixes.

Oliver asks if first iteration for fixes for POG based on feedback from lead user will be M5

    • Specifically, do we believe Harm's test case will work in M7
    • Alex believe benefits will start to be seen as early as M5 (by end of this delivery)

Will need another lead user as this one starts to finish up. Teams should be thinking of potential candidatates.

Trace items since Guru had to leave early

Question regarding resources for a few trace test buckets

  • There was a bugzilla asked to review -- 200955. AI to finish reviewing
  • A few of the tests are profiling cases covered by Oregon folks.
  • Others historically done by IBM team
    • For these, what will be done in I4 by Joanna's team
      • Joanna reports start in Jan for resource and helper
      • AI to Guru to send list of desired tests to Joanna
      • For M4 IBM will do what they can and remove others from test report.
  • List was sent to Joanna but lost in the mail; she now sees it.
    • Guru asks for response today. AI to Joanna to follow up with Guru

Question arose regarding whether the profiling tests should move to platform since that is where bulk of profiling stuff is being officially managed.

Project Housecleaning

Inactive committers list -- Has everyone sent out their mails?

  • We want minutes to reflect names of removed committers
  • AI to everyone send list to Chris of removed committers

Component reorganization

Trace/Platform -- n/a

Test Done

Monitoring -- work done to remove from GA. Paul asked if all work been done to put it on web site in "as is" section?

  • perfmon piece has not yet been put on asis.
  • log entry piece has been removed
  • Paul mention this needs to be done (also for rcp symptom editor/etc.)

Alex asks what exactly goes up on asis site... In particular, existing build or do we continue to build them?

  • Response: Whatever is easiest...
    • If not automatically building; post last good build there
    • If automatic building (for test components this is case), continue to build

Paul asks if it can this get done today?

  • Alex will need to look into what is done already
    • RCPs are ready from 4.4.0.3
  • Perfmon agent will take more work (2 parts client/server)
    • Alex to talk to Paul because paul had some 2 part stuff in manual test as well

Managing BTM between Monitoring and Trace

Trying to cloase an AI between AlexN and Guru

  • BTM trace execution component
    • Alex wants to track under trace
      • Motivation. code already in trace should be managed there and under monitoring managed there
    • Guru wants to track under monitoring
      • Motivation. Guru states "historically want each feature to require platform and its subproject not to require another subproject."

Discussed briefly and thought is that it should be managed where it falls within current matrix until we fix the matrix.

  • ARM agent of btm is in trace
  • ARM instrumentation of btm (in platform) -- functionally there is no trace.execution for ARM

Based on this heuristic, it is looking like component in question should be managed in Platform (based on matrix on web site)

  • There have not been discussions to change web site so it is considered correct.
  • Takeaway:
    • arm.agent not moving
    • arm.instrumentation (paul does not think should be in platform (although it is)... core underlying piece in platform...)
    • btm.wrapper in monitoring

AI for Joanna and AlexN to close on management for component in question.

More general question of component management

Questionarose whether we should invest a block of time at next F2F on component reorganization

  • Question about overall philosophy (as Guru suggests above or something else)
  • Tie in to potentially smaller discrete packagings

paul suggests -- telecon in new year to see if can quickly close before delaying it to f2f.

  • General agreement to follow this process.

= Test automation

Monitoring -- 3 or 4 test suites automated

  • Blocked by 2 AGR defects for addl test automation
    • On Alex' issues -- I believe Paul mentioned that he should talk to Duane

Platform -- Automated what planned to automate

  • Process documentation/strategy by end of week
    • Achieved 20% savings
    • Question: will we save more than invested... Yes

Trace/Profiling -- had some automated already

  • Remaining done by Joanna in Platform

Test

  • achieved all wanted except for 3 test cases that had AGR issues

AI for leads: Fill in numbers for reduced overhead of test cycles given automationthusfar

  • after BVTs we will want 3rd dataset regarding not having to run on reference platform

Over hump of automating

  • Some documentation is in process
  • Automatic report generation is proceeding.

4.5 I4

We are in test pass now. Hows it going?

Test

  • blocking bug on Monday fixed in latest candidate build
  • hope to finish this week.

Trace

  • Started Monday evening
  • plan to finish by next Tuesday (Intel part)


Platform

  • started yesterday
  • no issues forseen
  • automation is helping

officially TP1 ends on Tuesday (Dec 19). 3 days to fix before TP2. Schedule builds in about a week off (for western christmas break)

  • Beause of worldwide holiday schedules, Paul suggests finish TP1 fix stuff and get as much of TP2 done as possible before leaving for break.
  • Because of holiday schedule, everyone needs to keep eyes open as they choose what to fix in build for I4 in TP2.

MISC

AlexN will be out next wednesday.

AlexA raised issue with Russia holidays impacting the Intel part of the build process. between (Dec 31 to Jan 8)

  • Its all automatic (unless there is a problem... which does happen a decent percent of the time)
    • thru dec 31 -- Russia team is available to cover
    • Jan 1 everyone is gone anyhow
    • Jan 2, 3 -- Chris to cover. AI to Chris, Alex, Kiryl to ensure Chris is ramped.

Chris mentioned that he gave a Democamp presentation on the TPTP profiler at Portland Democamp. It seemed well received.

Back to the top