Skip to content

release: reduce nf_conntrack_tcp_timeout_syn_sent to 20s#907

Open
Mohijeet wants to merge 1 commit intobottlerocket-os:developfrom
Mohijeet:fix/reduce-syn-sent-conntrack-timeout
Open

release: reduce nf_conntrack_tcp_timeout_syn_sent to 20s#907
Mohijeet wants to merge 1 commit intobottlerocket-os:developfrom
Mohijeet:fix/reduce-syn-sent-conntrack-timeout

Conversation

@Mohijeet
Copy link
Copy Markdown

In clusters using VPC CNI, pod IPs can be reassigned within the default 30s cooldown after a pod crash. Stale SYN_SENT conntrack entries (120s default) outlast that window, allowing client retries with the same src IP:port to hit a different pod/service via the stale entry even after kube-proxy has updated routing rules.

Setting nf_conntrack_tcp_timeout_syn_sent=20 ensures entries expire before the minimum IP reuse window. Added alongside existing conntrack tuning in release-sysctl.conf.

@Mohijeet
Copy link
Copy Markdown
Author

more about issue : aws/amazon-vpc-cni-k8s#3634

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant