we do not support the use of TAP interfaces with ovs-dpdk and never have. The step descried in this doc including using udev rules to modify tap device are not supported in OSP the document also states " This is documented in the libvirt domain spec for KVM and can be done with virsh edit on a RHEL KVM hypervisor. " we do not suppoprt the use of virsh edit ever in openstack and if customer us it it will render there vm unsupported. the correct way to cange the tx/rx queue lenght is to set the tx_queue_size and rx_queue_size paramaters https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.rx_queue_size https://docs.openstack.org/nova/latest/configuration/config.html#libvirt.tx_queue_size Reported by: smooney https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html/ovs-dpdk_end_to_end_troubleshooting_guide/high_packet_loss_in_the_tx_queue_of_the_instance_s_tap_interface#annotations:5767a912-9b85-4719-9d00-c34a848290c2
it looks like that incorrect info may have orginally came form https://access.redhat.com/solutions/2785881 it would have worked for none ovs-dpdk environment before we added the capablity to manage queue lenght via nova but this type of turning of tap devices outside of openstack is not something we should be supporting in OSP today. let ensure that https://access.redhat.com/solutions/2785881 is also removed.
The KCS https://access.redhat.com/solutions/2785881 has been updated to remove the unsupported work around for modifying qlen for via udev. That section has been replaced with a link to "Change RX queue size and TX queue size of virtio NICs deployed with Red Hat OpenStack Director" at https://access.redhat.com/solutions/3641441.
updates published at https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/13/html-single/ovs-dpdk_end_to_end_troubleshooting_guide/index#high_packet_loss_in_the_tx_queue_of_the_instance_s_tap_interface