Adrian Moreno <amore...@redhat.com> writes: > On 11/16/23 18:59, Ashish Varma via discuss wrote: >> Hello all, >> I saw a setup where the ct action was used like: >> ct(table=100, commit, zone=6000, >> exec(move:NXM_OF_IN_PORT[0..15]->NXM_NX_CT_LABEL[48..63])) >> According to the ovs-actions man page: >> " >> Without*commit*, the*ct *action accepts the following arguments: >> *table=*/table/ >> Sets the OpenFlow table where the packet is reinjected. >> The/table/ must be a number between 0 and 254 inclusive, >> or a table’s name. If/table/ is not specified, then the >> packet is not reinjected. >> " >> It seems it is wrong to put both commit and table in the same ct action. >> What would be the behavior when the packet matches the flow with this action? >> It seems to be going on to table 100. >> Thanks, >> Ashish. >> > > Hi Ashish. > > You're right. It's confusing. OVS is not enforcing the rules that the > man page describe. I was curious and looked around and found that the > dpdk datapath even forces the commit flag to be present in nat actions > [1]. > > Maybe Aaron and Paolo have more context?
There may be some issues in the ovs-actions manpage as you note. Conntrack action changed quite a bit from inception to implementation to where it is today. I guess this section of the manpage should be rewritten. > [1] > https://github.com/openvswitch/ovs/blob/8b5fe2dc6080db0bed9969cf81bb4a007539cfbe/lib/dpif-netdev.c#L9229-L9232 > > >> _______________________________________________ >> discuss mailing list >> disc...@openvswitch.org >> https://mail.openvswitch.org/mailman/listinfo/ovs-discuss _______________________________________________ discuss mailing list disc...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
