Page History

›How can we release more frequently, smaller scope without impacting the SW quality?

›How can we change the waterfall-ish development model?

›How can we get enough details about candidates requirements prior the release?

›How to manage the release in case of people turn-over?

›How to handle the scope and maintain a balance between Usecase/Functional reqs (attractive) and Non functional reqs (less sexy)?

•   Identify a set of projects that could prototype the Release Train approach
•  

   “Release cadence transition proposal” on June 25th, 2020 @ 11am UTC

•   Alignment with CNTT/CNCF TUG and O-RAN/ONAP in order to define their role as part of XGVela and identify what can be leveraged from the existing projects
•  Introduction of XGVela to the TAC team

Track: Cloud Native in Telecom Cloud

Presenters/Moderators: Qihui Zhao, Ying Li

•  Alignment with CNTT/CNCF TUG and O-RAN/ONAP in order to define their role as part of XGVela and identify what can be leveraged from the existing projects
•  Introduction of XGVela to the TAC team

Reviewed mandatory upgrades to Java 11 and Python 3 for all projects

Migration of ONAP to standard infrastructure versions: Docker, Kubernetes, OS images, databases, etc (see Database, Java, Python, Docker, Kubernetes, and Image Versions)

Updating vulnerable direct dependencies:

• SECCOM repo specific recommendations on the Security Vulnerabilities protected wiki space
• Must be complete by M2

Resources to do the non-functional work

Potential dependencies that conflict with new versions

SECCOM will have representative at the weekly PTL call to answer questions

PTLS and others are always welcome to attend the SECCOM weekly on Tuesdays to raise issues

PTLs secure resources and plan to identify show stoppers as soon as possible (M1)

EUAG Operator Survey & In-Depth Analysis On Consumption Model

Presenters: Atul Purohit

• EUAG should create an action plan out of survey inference, what it means for various committees and how can the feedback be provided back to CSPs
• Survey sample was about 50% of overall members and 75% of active members, to make similar activities more impactful in future perhaps the survey can be done with larger sample size

OOM Status update: what's planned internally

Presenters/Moderators: Sylvain Desbureaux, Krzysztof Opasiak

Review of the planned OOM changes :

Support of Helm V3

Kubernetes V1.17 (or 1.18)

Migrate to Seccom recommended Versions

Update defaults (use Ingress, Storage Class, Hardened OS)

OOM Status update: consequences on other components

Presenters/Moderators: Sylvain Desbureaux, Krzysztof Opasiak

Review of the required changes on components helm charts VS requirements

presentation of changes, use of templates, adding appender to Logback to support STDOUT as additional output

Container that do not contain ONAP code should not be hosted on Nexus

No Root access to DB

Application config should be fully prepared before starting the container

Containers must crash properly when a failure occurs

No more Nodeports

AAF optional (component should work without AAF even in degraded mode)

HTTPS is mandatory but should be configurable (disable in case of Service Mesh as this will be offloaded to Service Mesh)

Container RootFS should be mounted as ReadOnly

Commit message rules for OOM

Requirements Traceability:  Initial Request through TSC Approval

Presenters/Moderators: Alla Goldner, Chaker Al-Hakim, Pawel Pawlak, Pamela Dragosh, David McBride, Catherine Lefevre

• brainstorming and discussion on aligning the ONAP requirements pipeline 
• Consensus:
  • any requirements- regardless of source- should go through the requirements subcommittee.
  • and euag submits requirements to req. sub where they are vetted
  • committee consolidates backlog
  • committee should make prioritization recommendations to the TSC

wiki: 2020 June vDTF ONAP Requirements Traceability: Initial Request through TSC Approval

• How can we streamline the requirements coming from different sources inside and outside from the ONAP Community?
• As an example, EUAG → REq Subcommittee prioritirised → (Architecture Review) → TSC 

• Enhance the mission of the Requirements Subcommittee:
  
  • Act as the ONAP Product Owners 
  • Recommend Prioritization to the TSC
  • Create the consolidated ONAP Backlog

The presentation covers Frankfurt CNF instantiation improvements on vFW use case example

• Changed modelling of the vFW CNF - split into 4 helm packages to benefit from CDS resource assignment
• Change from a'la Carte VNF-API instantiation flow into Macro GR-API with CDS
• Utilization of CDS for automatic assignment of Helm package overrides
• CDS uploads optionally profile which allows for further helm enrichment like extra k8s resources

Use Case Doc: https://onap-doc.readthedocs.io/projects/onap-integration/en/latest/docs_vFW_CNF_CDS.html#docs-vfw-cnf-cds

Notes: 2020 June vDTF ONAP vFW CNF use case evolution

• We lack of the use case automation scripts, however, there is very good documentation + postman collection
• We leverage VNF flow in SO and still, we need to use dummy heat templates in the onboarding package
• vFW use case requires dedicated k8s cluster with virtlet, ovn4k8s and multus
• Data in AAI is still not synchronized with k8s
• Status of instantiated resources is not monitored by SO and

• vFW CNF Use case automation with robot scripts and use of modern VID UI
• Support of Close Loop
• https://jira.onap.org/browse/REQ-341 - CNF SO orchestration Enhancements
• Potentially new use case: VNF + CNF Heterogeneous service and/or pure CNF (without a need of specific k8s cluster setup)

• Did use SDC and MultiCloud K8S plugin, but did not use SO.  Did have one manual step.
• Container image was in a local K8S repository, not bundled in the service package

• Add handling for the macro flow
• migrate onap_tests repository to pythonsdk_tests
• Add close loop examples coming in next versions

Release more frequently VS time

•  Move to a more continous approach
•  Release at each milestone ? with working small steps
•  At RC0 provide %age completed, which ones are not done yet?

The code (as demonstrated) along with test scripts and swagger documents will be added to the multicloud/k8s repository soon.

https://github.com/onap/multicloud-k8s

• Continue discussion in CVC.
• Map 4-tier structure to CNTT.

• Interesting and detailed presentation by Ying Li and Shasha Guo of SDN implementations for several use cases like Traffic Mirroring and Routing Configuration, showing code snipets, network diagrams and parameter mappings

• Presentation on CNTT OpenStack Release selection by Pankaj Goyal et al, and second half by Shasha Guo on Cyborg Acceleration Mgmt.
• OSTK Pike was selected in Paris, but the next version was selected by a formal process and against defined criteria, the process started at Baldy vF2F in April 2020.  Train met the selection criteria and is recommended as the next CNTT OSTK version. Recommendation: utilize Ussuri OSTK release for Cyborg API v2.0 service (as it is incompatible with Train). Upstream: Cyborg should fix their API in Train release as per OpenStack policy.
• Discussion whether CNTT should jump straight to Ussuri to avoid back porting problems for Cyborg. An option for consideration for TSC (the only criterion not met by Ussuri is 6+ month requirement)
• Shasha Guo and Ying Li on Cyborg acceleration: presented scenarios why we need to use Cyborg, and Cyborg enhancement requirements for CNTT.

ODL Micro Status & Next Steps

• Slide Deck here - https://docs.google.com/presentation/d/1hqN9cFzmzkafCPgcEx7wZFB-rPaWHvKOBitcElNUYwk/edit?usp=sharing
• Need performance tests that show the improvement by ODL-Micro v/s OSGi/karaf
• Need performance data about how much percent improvement ODL-Micro gives when testing with a device

•  Tejas Nevrekarto share performance reports once available. Further elaboration in TWS once code is uploaded

ODL Platform API Changes and impact to downstream consumers