Openshift: dropped packets over vxlan due to nf_conntrack race condition
Issue
- 200ms delay and TCP retransmits seen in Openshift environment when POD is transmitting to another node
- A slight delay is sometimes experienced when connecting to a POD over TCP or when it services some app-level request.
- Dropped UDP packets when sending UDP packets over vxlan.
Environment
- Red Hat Enterprise Linux 7
- Red Hat OpenShift
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.