Hi Morgan,

On Mon, Jul  8, 2019 at 03:47 PM, Morgan Richomme wrote:
> On our side we used to install incrementally ONAP using 4 steps
> - core (6): A&AI, DMAAP Message Router, Portal, SDC, SDNC, SO
> - small (9): core + AAF, APPC, CLI, External API, Log, MultiCloud, VID
> - medium (14): small + CLAMP, DCAE, Holmes, OOF, Policy, UseCase UI
> - full (61): all the components (VFC + all VFC drivers, VNFSDK, Modeling,
> Pomba, Multicloud drivers, Dmaap Bus controller, ...)

Does the "install" phase for these pods actually result in the pods starting up 
as well?
Or are the pods only started after everything has been "installed"?

AAI depends on AAF during startup.
AAI-Resources deployment chart contains MSB configuration that needs to be 
injected during startup.
So, AAF and MSB should be started before AAI, but neither of them appears in 
the "core" list.
For that matter, the common Cassandra should be in the "core" list too.

This installation sequence seems different from the proper startup sequence.

Doing these 4 steps appears to be explicitly countermanding the Helm chart 
dependencies, i.e. the "initContainers" section of the deployment.yaml files.
If the Helm chart dependencies are not properly expressed, then tweaking some 
timers is not a guaranteed fix for other operators who run ONAP in their 
environments.

In the near future, the intention is for AAI pods to generate AAF certificates 
on startup (as demonstrated in the aaf-hello pod), so this particular sequence 
seems likely to fail again.


Keong

-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.

View/Reply Online (#17920): https://lists.onap.org/g/onap-discuss/message/17920
Mute This Topic: https://lists.onap.org/mt/32393410/21656
Group Owner: [email protected]
Unsubscribe: https://lists.onap.org/g/onap-discuss/unsub  
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to