OOM Meeting Notes - 2017-12-06

Owned by Michael O'Brien
Last updated: Dec 08, 2017
3 min read

Agenda

  • @Alexis de Talhouët stabilization of the release-1.1.0 branch before KubCon and the F2F last thu/friday

  • @Alexis de Talhouët assisting and finishing the vFW to vf-module creation + video + wiki details last friday

  • TSC discuss committer responsibilities

  • TSC discuss dormant committers and making room for new committers

  • Discuss how to do integration testing on pending reviews

  • Add all committers/contributors to reviews - and give enough time to review

  • Please commit the configuration (mso prototype) changes 

  • Discuss tosca commits and reviews - expanding review - ie: https://gerrit.onap.org/r/#/c/24719/ 

  • F2F discussion

  • KubeCon demo yesterday - their environment was good as of Monday

  • JIRA board

  • Meeting next week (8AM PST) ? confirm

  • Borislav's raised in the Tue Logging meeting - logging filebeat issue onap/ecomp + TBD in recent commits - LOG Meeting Minutes 2017-12-05

    • check on lost logging configuration on the logging-analytics repo moved to oom 

    • specifically docker/src in the config container

    • Fix for current 1.1.0 

    • mso jetty container inside the war (logger) has issues unless put in after war explode - verify post start hook

  • Getting ready for ONAP dev event:

    • Document list of OOM requirements for Components teams to achieve carrier grade

  • CI/CD for OOM

    • We need to establish a good CI process for OOM to avoid regression impacts to other projects. 

    • @Alexis de Talhouët has a proposal on how to do it

    • @Michael O'Brien - (deprecated as of 20170508) - use obrienlabs has a CD script 

  • Creation of an Amterdam branch

    • @Alexis de Talhouët will create the amsterdam branch out of the last commit.

Analysis of the onap/ecomp issue in MSO with the ELK stack raised by Borislav

Analysis of the onap/ecomp issue in MSO with the ELK stack raised by Borislav

A lot of the fixes since Dec 30 were done during the 3 day blitz to stabilize ONAP in prep of KubeCon and the F2F - during validating the vFW

Message: Regression test – even though MSO flows did not work until the change – we can still run rest calls to force out logs



If you look at some of the merges – we might get hammered on being light on reviewers for master – but you would need to know this was a cherry picked form 1.1 where the fixes were done first - and tested on 2 systems –which does have all the reviewers. The issue is a lack of regression testing for logs.

We need to have a plan to address changes that are approved but still cause design contention and are raised later – in the logging meeting the issue is phrased more of a design refactor issue not something that is actually fully broken - again need to retest the MSO logs - which we can now that vf-module creation works because of @Alexis de Talhouët

https://gerrit.onap.org/r/#/q/status:merged+project:+oom

For example https://gerrit.onap.org/r/#/c/25053/

was flagged as busted – but reviewed as ok because a separate patch below reverted the onap/onap back to onap/ecomp for filebeat/logback

the fix https://gerrit.onap.org/r/#/c/25219/1

fixed this issue and was signed off – if we are talking about another change then we may need to create a new JIRA for it when we identify it

I will add an agenda item to the OOM meeting tomorrow to get this ongoing issue resolved – as in identify exactly the issue in a JIRA and add all reviewers before we adjust the yamls again –and get Borislav’s signoff.

Which brings up the fact that we likely need to consume public onap daily and run a log sanity on it asap – especially after a large number of commits and between the port to master.

We need a process for this before being put into integration testing after the F2F.

  • additional





Notes

Next week

Alexis cherry picked a fix from the AAI team that exposed a master to release-1.1.0 issue (we now point at AAI amsterdam) - was broken for 8 hours - traversal and resources are good again OOM-501: AAI resources failure around 1000EDT on release-1.1.0 - check AAI-548 cassandra env config changes to portClosed