Istio CNI - memory spikes and OOM errors #48922
Unanswered
AleksanderBrzozowski
asked this question in
Q&A
Replies: 1 comment 8 replies
-
there might be room for improvement here on the memory usage but increasing the limit seems like an obvious short term step to make an improvement |
Beta Was this translation helpful? Give feedback.
8 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hi, we have a EKS Kubernetes cluster in AWS with Istio mesh, version 1.17.8.
We use Istio CNI to configure networking, and we observe memory spikes every ~ 50 minutes:
We figured out that it is caused by service account token refresh, but what we don't understand is why memory increases suddenly? These memory spikes sometimes cause OOM errors. The same memory spike we observe during startup.
We tried upgrading Istio to 1.19.6, but we observed the same behavior after upgrade. Furthermore, setting
resources.memory.limits
of the Istio CNI deployment to100Mi
causes it to fail instantly with OOM.Here are Istio CNI logs when service account is refreshed (from version 1.19.6):
Resources:
Container settings:
The question is - should we just increase
memory.limits
of Istio CNI pods to avoid OOM errors, or maybe there is something wrong with our setup?Beta Was this translation helpful? Give feedback.
All reactions