- Github renewal
- Jira status
- Review PDP-12 – Cloud Native PNDA
- Red PNDA for ONAP?
- What is required for Kube deployment of Red PNDA?
- Do we have interested contributors?
- Casey Cain
- Frank Brockners
- Alexei Nekrassov
- Borislav Glozman
- Carlos Giraldo
- Cristina Precup
- Kiran Ambardekar
- Vijay Venkatesh Kumar
- Vincent Colas
Frank Brockners will take renewal issue offline with Casey.
Donald gave a readout of the current Jira status:
- Jira reports 410 unresolved issues, but this includes items marked wont-fix. There are 349 open issues.
- 17 Epics – https://issues.pnda.io/browse/PNDA-4552?filter=10502
- 59 Features – https://issues.pnda.io/browse/PNDA-4919?filter=10501
- 69 Improvements – https://issues.pnda.io/browse/PNDA-4913?filter=10505
- 56 Bugs – https://issues.pnda.io/browse/PNDA-4938?filter=10500
- 109 Tasks – https://issues.pnda.io/browse/PNDA-4264?filter=10504
Donald has started reviewing open issues to review priority, add evaluation comments or mark wont fix.
Review of PDP-12
Carlos gave a readout about the information he added to PDP-12: Cloud-native PNDA. There were several points that he highlighted:
- They would pick an official upstream container / helm chart if one was available.
- They have built their own Dockerfile for some components when no official one was available, or the unofficial one was not a good fit.
- Several PNDA components use different value formats for the same customisation parameter. For example, a mixture of hostname, URL, etc. gets used to identify the hadoop-name-node. Need to clean up PNDA components to use a consistent set of parameters.
- Need to standardise on when to use ENV parameters versus inject config files into containers.
- Some PNDA components, e.g. deployment-manager use shell commands and systemd scripts to deploy applications. This needs rethought to be cloud-native.
Red PNDA for ONAP?
Vijay described his view of what it would be nice to see from an ONAP / DCAE integration perspective. A fully containerised PNDA such as Red PNDA integrated into Kubernetes is desirable but Vijay is more interested in integration with policy and orchestration for DCAE applications. We discussed the edge cloud orchestration with Helm versus the DCAE orchestration with Cloudify and the lack of compelling use cases to drive PNDA integration. It is really not clear whether an embedded PNDA instance will be needed at the edge or centrally, or indeed whether deployments will want to integrate with an external big data platform. We discussed the need for a use-case other than TCA to help define the analytics deployment requirements.
Donald said he would take a fresh look at application integration to find an approach that could work for the different deployment scenarios.