Hi ... I know that we need handoff mechanism when running multithread because traffic for specific inside network user must be processed always on same thread in both directions and We can not remove handoff node from NAT nodes ,because handoff is faster than locking mechanism. So the problem is potential dead lock when two workers waiting for each other!! for example Worker A and B, A is going to handoff to B but unfortunately at the same time B has the same thing to do to A, then they are both waiting forever. your solution In VPP 19.01 (or 18.10) is dropping packets when the queue is full (congestion drop)
first question is how you can detect congestion on queue ? and What happen if we have separate CPUs for Handoff node and nat processing node and do not the same job handoff nad natting on single CPU core. for example CPU core A nodes are ( dpdk-input -> ... -> NAT handoff ) and CPU core B nodes are ( nat44-in2out -> ... -> ip4-lookup -> interface-output) , in this situation worker A wait for worker B but worker B never wait for worker A. Is it true to say we never have potential dead lock if we have seprate CPUs ? If yes, why you use a same single CPU for both nat and handoff ? The separate CPUs can not solve the deadlock ? what happen if we have seprated CPU core for Handoff nodes and nat processing and do not the same job handoff nad natiing on single CPU core. for exapmle CPU core A nodes are ( dpdk-input -> ... -> NAT handoff ) and CPU core B nodes are ( nat44-in2out -> ... -> ip4-lookup -> interface-output) , in this situation worker A wait for worker B but worker B never wait for worker A. so Is it true to say we never have potential dead lock if we have seprated CPU ??? If yes, why you use a cpu for both nat and handoff ? seprated CPU can not solve the potential dead lock ?
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#11794): https://lists.fd.io/g/vpp-dev/message/11794 Mute This Topic: https://lists.fd.io/mt/28878687/21656 Group Owner: vpp-dev+ow...@lists.fd.io Unsubscribe: https://lists.fd.io/g/vpp-dev/unsub [arch...@mail-archive.com] -=-=-=-=-=-=-=-=-=-=-=-