-
Bug
-
Resolution: Done
-
Medium
-
Dublin Release
-
OOM-Staging-Daily
-
Integration R5 Test1(08/19)
Many pods fail to come up during ONAP deployment. It appears CPU usage on Rancher node and cluster nodes are very high and it causes pod liveness and readiness probe fail.
The attachments are CPU usage on failing Rancher node (deployed on 03/22/19) and normal Rancher node (deployed on 03/14/19). The high CPU one is from the failed ONAP deployment.
The cluster node CPU is also high:
root@oom-rancher:~# kubectl top node NAME CPU(cores) CPU% MEMORY(bytes) MEMORY% oom-etcd-1 1442m 72% 1677Mi 43% oom-etcd-2 1538m 76% 1406Mi 36% oom-etcd-3 1426m 71% 1425Mi 37% oom-k8s-01 5366m 67% 9787Mi 61% oom-k8s-02 7468m 93% 9509Mi 59% oom-k8s-03 5726m 71% 4992Mi 31% oom-k8s-04 7644m 95% 10544Mi 66% oom-k8s-05 5316m 66% 7484Mi 46% oom-k8s-06 7174m 89% 7833Mi 49% oom-k8s-07 6912m 86% 12345Mi 77% oom-k8s-08 6421m 80% 12206Mi 76% oom-k8s-09 7424m 92% 13497Mi 84% oom-k8s-10 2687m 33% 7067Mi 44% oom-k8s-11 6544m 81% 6508Mi 40% oom-k8s-12 5744m 71% 9836Mi 61% oom-orch-1 1420m 71% 987Mi 25% oom-orch-2 1859m 92% 2395Mi 62%
- is blocked by
-
-
- Closed
-
- links to
