Gary, In working with OOM over the last few week it seem like the best way to detect problems is to run the demo-k8s.sh init and then to check via op0001 if SO, AAI and SDNC picked up the models. Usually on a fresh OOM they do not. The repair is to delete the AAI model-loader POD, the SO POD and the UEB-Listener POD in SDNC (I usually do dmaap-listener as well). K8 will restart those PODs. I then do a redistribute in SDC as op001 and see success. Every once in a while I also have to delete the sdc-be/sdc-fe to repeat the deletes if the sdc-be wasnt up properly (usually because I didnt check healtcheck first on sdc)
I'm wondier if this is a CSIT jenkins job we should do to run init and instantiate VFW after healthcheck ? Or just a custom check after 'demo-k8s.sh init' or 'demo-k8s.sh onap init' to test that the distribution actually succeeded to demonstrate installation health ? The distribute data can be retrieve by the catalog services API /sdc2/rest/v1/catalog/services/1eb7b5c8-9ec9-4090-8bef-322d82f5400c/distribution) where the 1eb... is the catalog-service-uuid to get the distribution ID and then /sdc2/rest/v1/catalog/services/distribution/6cf76bba-62d7-4dcb-8410-6760fc54cee7 to get the table of the actual status for the notified components Brian
_______________________________________________ onap-discuss mailing list [email protected] https://lists.onap.org/mailman/listinfo/onap-discuss
