The following projects are requesting waivers from the Dublin Sonar
coverage goals:
SDC
AAF
Multi-cloud
-=-=-=SDC Plan - May 16th email-=-=-=-=
Dublin code is covered, so removing it won’t help here. Some complex legacy
code is still hard to reach with unit testing, and that’s the part that
still isn’t covered.
We miss about 6%, hence we must cover about 15,000 lines.
My plan is to encourage even more coverage on new code in the first place,
so it would balance;
and to refactor some legacy code in El-Alto which can be covered more
easily comparing to legacy.
Based on current tests contribution rate and giving that plan, we will
hopefully reach the goal by Frankfurt sign-off.
-=-=-=-=-=AAF plan - May 9th email-=-=-=-=-=-
AAF officially requests Test Coverage Waiver.
Facts:
- WHEN AAF was created (NOT in ONAP), AAF used JUnits only for Classes
that are reasonably Unit Testable.
- MOST of the testing was done in a different Testing Scenario
- MOST of the important testing of AAF is actually around Identities
coming in on Transactions, which is not very
- AAF consists of a larger code set than many services. It involves 6
standard Services, encompassing elements of Security
- AAF has high compliance marks on everything required in ONAP, except
this one benchmark.
- AAF was brought in for Security.
- AAF’s Benchmark was moved to 65% along with everyone else’s.
- All my time, as most expert AAF developer, was spent on critical
dynamic Certificate creation strategy.
- AAF declared at the beginning of Dublin that we have few resources
- only 1 Person time assigned to ONAP as a whole from our Company
- there is no one from the community trying to come in and help with
JUnits.
- There is some help with working through SONAR “Errors” and other
technical debt, which is appreciated.
Quality:
- Junit usage is typically touted to increase product quality, and avoid
Regression Mistakes
- As stated, AAF code set has demonstrable code quality
- AAF had Zero Downtime in AT&T for 2018, and 3 mins of degraded
minutes for a particular app in 2017
- This has garnered praise from our leadership for high Architectural
and Implementation Quality.
- THIS is the code set quality that was brought in *gratis* for ONAP.
- As stated, we used an internally developed Regression Tool to
accomplish this, not JUnits.
Risk/Benefit:
- There are absolutely no Benefits of going with older Cassablanca code
over Dublin code.
- Improvements in OOM stability alone is reason enough to keep the
Dublin code over Casablanca.
- To remove AAF at this point for Security does NOT increase Security,
but exposes ONAP as being practically “Security-less”.
Plans forward:
- It was critical for AAF to develop a plan for K8s dynamic creation of
Configurations and Certificates for ONAP
- This is accomplished in Dublin, but not ready in time to get ONAP
services themselves to utilize… That is to be expected for Infrastructure.
- Most of El Alto needs will be in helping Clients use this
capability.
- I am a realist.
- The probability of QUALITY help for AAF in or out of ONAP
approaches 0% in 2019.
- What can be done must, essentially, be done by myself, with some
help by Sai G.
- I am not assigned ONAP for more than 50%, as I still am the main
person within my Company for this successful, but critical service.
- I propose that for El Alto, AAF (meaning me) focus ONLY on
- Helping ONAP Apps to utilize dynamic Cert strategy
- Fixing anything around this necessary (i.e. new requirements that
show up)
- Junit increase – I will take direct Developer action, which should
speed up.
- Only the Required ONAP System changes, examples:
- Move of OOM to AAF Repos
- Minimize other improvements… perhaps to One, i.e. Logging
- Without additional quality resources, the expectation is to provide
SOLID Improvement release to release.
-=-=-=-=-=-=-=Multi-Cloud Plan - May 8th email-=-=-=-=-=-=-=
I would like to request a waiver from the TSC in Dublin release for
following multicloud submodule :
https://sonar.onap.org/dashboard?id=org.onap.multicloud
.azure%3Amulticloud-azure
Facts:
MultiCloud project consists of a framework service and multiple plugin
services. While framework service are essential for multicloud’s function,
all plugin services can be independently deployed without any concern
compromising the multicloud functionalities.
Multilcoud-azure is one of plugin services which is to adapt Azure cloud to
ONAP for VNF orchestration. This plugin has been proposed and contributed
by Sudhakar from Amdocs, and it had been meeting the Sonar Goal in
Casablanca Release (51.5% > 50% goal).
There is no any new functionality added to this plugin service, neither
increasing UT coverage to meet the new Sonar Goal (55%).
Quality:
Have a deep dive into the source code of multicloud azure
plugin,
https://sonar.onap.org/component_measures?id=org.onap.multicloud.azure%3Amulticloud-azure&metric=coverage
I can see that the existing UT codes covers most of critical
functionalities. Those not covered are functionality not yet utilized yet,
or settings file, swagger, etc.
So I believe the quality of source code is still in good
shape.
Risk/Benefit:
The risk is that there is lack of commit resource to support
this plugin service for Dublin, not clear for future releases. However, As
some service providers expressed their interests to utilize this plugin
soon, I believe there will be resource to join and support it in future.
Plan forward:
We will call for the help to support this multicloud-azure
plugin in F-release, otherwise, we might put it as inactive status, I mean,
it will not be part of future releases.
--
Jim Baker
Linux Foundation Networking - Technical Program Manager
mobile: +1 970 227 6007
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#5005): https://lists.onap.org/g/onap-tsc/message/5005
Mute This Topic: https://lists.onap.org/mt/31730526/21656
Group Owner: [email protected]
Unsubscribe: https://lists.onap.org/g/onap-tsc/leave/2743226/1412191262/xyzzy
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
