>-----Original Message----- >From: Advith Nagappa [mailto:advith.naga...@gmail.com] >Sent: Monday, March 27, 2017 9:45 AM >To: Bodireddy, Bhanuprakash <bhanuprakash.bodire...@intel.com> >Subject: Re: [ovs-discuss] Userspace OVS - Kernel Hashtable > >Hi, > >Thank you for the response! > >Is there a way to ensure that particular flow rule is never cached? (i.e, to >translate the action for each flow-rule hit)
Please don’t drop the list. In the master, the default behavior is not to insert every flow in to EMC. If a flow is hit 100 times It gets inserted in to EMC. This is due to probabilistic insertion behavior that was recently introduced. You can find more information and the options to tune in section "EMC insertion probability". http://docs.openvswitch.org/en/latest/howto/dpdk/. - Bhanuprakash. > >On Mon, Mar 27, 2017 at 10:31 AM, Bodireddy, Bhanuprakash ><bhanuprakash.bodire...@intel.com> wrote: >>-----Original Message----- >>From: ovs-discuss-boun...@openvswitch.org [mailto:ovs-discuss- >>boun...@openvswitch.org] On Behalf Of Advith Nagappa >>Sent: Monday, March 27, 2017 7:03 AM >>To: disc...@openvswitch.org >>Subject: [ovs-discuss] Userspace OVS - Kernel Hashtable >> >>Hello All, >> >>Does a purely userspace OVS bridge (kernel module not installed) also use >the >>Kernel Hashtables to cache flows? >> >>If not, how does userspace OVS cache flows (if it all it does?) > >There exists this Exact Match Cache(EMC) per PMD thread and has 8k entries >to cache flows. In case of lookup failure >the packets are sent to datapath classifier. The dpcls can have multiple >subtables and key lookup is done on each subtable >till a match is found. Note that there exists a classifier per ingress port >in OvS- >DPDK master! > >- Bhanuprakash. _______________________________________________ discuss mailing list disc...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
