rx-miss typically means that you are not pulling packets out of queue fast enough, as probably your CPU is too busy with processing packets. modifying queue size will not help you with constant traffic rate, it may help only with bursty traffic and in cases when VPP is de-scheduled by kernel.
— Damjan > On 11.11.2020., at 11:13, Merve <merveorakc...@gmail.com> wrote: > > Hi everyone, > when the rx and tx queue number is low, dpdk catches the packets faster, but > when the rx and tx queue number is high, I get a result like, below. I have > rx-no-buf and rx-miss. So in both cases I have packet losses. how to get rid > of this situation... > > > > startup.conf > > dev default { > num-rx-queues 2 > num-tx-queues 2 > num-rx-desc 1024 > num-tx-desc 1024 > } > > > Name Idx State MTU > (L3/IP4/IP6/MPLS) Counter Count > TenGigabitEthernet1/0/1 2 up 9000/0/0/0 rx > packets 34285615 > rx bytes > 2057136900 > tx > packets 34285610 > tx bytes > 1577138074 > drops > 5 > ip4 > 34285609 > rx-miss > 222380741 > tx-error > 4 > > > > > ************************************************************************************************************** > > > > startup.conf > > dev default { > num-rx-queues 8 > num-tx-queues 8 > num-rx-desc 1024 > num-tx-desc 1024 > } > > > > Name Idx State MTU > (L3/IP4/IP6/MPLS) Counter Count > TenGigabitEthernet1/0/1 2 up 9000/0/0/0 rx > packets 7705 > rx bytes > 462300 > tx > packets 7702 > tx bytes > 354306 > drops > 3 > ip4 > 7701 > rx-no-buf > 768530880 > rx-miss > 204257196 > tx-error > 2 > > > > thanks for your help! > >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#17987): https://lists.fd.io/g/vpp-dev/message/17987 Mute This Topic: https://lists.fd.io/mt/78179815/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-