On 7/9/21 6:11 PM, Han Zhou wrote: >> To avoid this potentially expensive table walk, we use the load_balancer >> uuids stored in the datapath record itself (it's probably best to see >> those as hints I guess). >> > Thanks for the explain. What you described is indeed a dependency between > lflow and sb_load_balancer because in lflow's compute/change handlers > sb_load_balancer data is required. (otherwise we would not need to call > sbrec_load_balancer_get_for_uuid(). > > However, since this dependency is already captured in the I-P, it is just > easy for this use case. We should simply use > sbrec_load_balancer_table_get_for_uuid() instead, which takes struct > sbrec_load_balancer_table* as argument and we already have it in the > lflow_ctx_in.lb_table as the input to lflow engine node. >
You're right, it's simpler like this, thanks for pointing out the sbrec_*table_get_for_uuid() variant. I sent a v2: http://patchwork.ozlabs.org/project/ovn/list/?series=253029 Regards, Dumitru _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
