-
Notifications
You must be signed in to change notification settings - Fork 1.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Calico's performance is not as expected and seems unstable on my environment. #8756
Comments
Do you use vxlan? Are the nodes in different subnets? I suppose there is offloading turned off on |
@tomastigera Yes, all nodes are connected to the same switch and VXLAN is not being used. I have set the encapsulation to None. I found in subsequent testing that the total bandwidth can reach the expected value when using concurrency parameters(-P 10), but it is still only about half of the physical bandwidth in single-threaded scenarios. I also tried using ethtool to disable rx-checksumming and tx-checksumming, but I didn't see any significant change. I also tried starting two iperf3 containers on the same node to test the same target simultaneously. I expected the results of the two iperf3 containers to add up to the physical bandwidth. However, in fact, the test results of each iperf3 were lower. I can't think of where the problem might be. I thought that in the case of "encapsulation: None", calico only needs to maintain the local routing table and veth pair, and there should not be such a large difference with the physical network. |
Expected Behavior
Under normal circumstances, the network performance loss of Calico should be within 10%, right?
Current Behavior
Calico's performance is not as expected and seems unstable. How can I further analyze the cause of network degradation and solve this problem?
The network loss from testing on nodes to pods is over 50%!!!
Possible Solution
Have I missed any important system kernel parameters or Calico configurations?
Steps to Reproduce (for bugs)
Context
1、node to node:
2、pod to node:
3、node to pod:
4、pod to pod:
Your Environment
The text was updated successfully, but these errors were encountered: