Thanks for your reply, but how does the userspace make the decision about flow table clear? As far as I know, OVS use this function u64 ovs_flow_used_time() to calculate the time flow is used, and then submit to userspace in statistics with function static int ovs_flow_cmd_fill_info(). Can you show me the source file about decision making process or some function about it, if that is possible, Thank you very much. --Binghan
On 2012-4-4, at 下午5:42, Justin Pettit wrote: > On Apr 4, 2012, at 8:36 AM, LIU Binghan wrote: > >> I think there are two possible ways it could be. >> >> First, in the beginning, datapath does not have any flow table. If it found >> a flow that it does not know, then it would pass the flow key and packet to >> vswitchd. Vswitchd processes the packet, and also tells the datapath to >> expand its flow table about this flow. After that, packets would process by >> kernel. And then, if after several seconds some flow does not show up, the >> flow table will delete the flow from it. >> >> Second, datapath just utilize some exist flow table in the kernel(not kernel >> module) and only learn the flow that they do not know from vswtich. >> >> So which one is right or both them are wrong? Thanks for your answer. > > The first one. The only correction is that userspace makes the determination > on when the kernel flow table should be cleared, not the kernel module on its > own. > > --Justin > >
_______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
