On 2/28/25 5:07 PM, Dumitru Ceara wrote: > On 2/27/25 9:58 PM, Lorenzo Bianconi wrote: >> Introduce the capability to actively refresh mac_binding entries available >> in OFTABLE_MAC_BINDING table when they are inactive for more than the >> configured threshold. >> >> Changes in v5: >> - Reduce threshold dump_period and cooldown_period to 3/16 of threshold value >> in order to not miss any re-arping time slot. >> - improve unit-test and system-test >> >> Changes in v4: >> - Set req_delay to min(dump_period) / 2 in mac_binding_probe_stats_run() >> - Fix MAC binding aging - probing GW router Dynamic Neigh unit-test >> >> Changes in v3: >> - add system-test >> - modify re-arping timeslot boundaries >> >> Changes in v2: >> - add MAC binding aging - probing GW router Dynamic Neigh test >> - improve unit-tests relying on fmt_pkt() >> >> Reported-at: https://issues.redhat.com/browse/FDP-1135 >> >> Lorenzo Bianconi (3): >> controller: Update OFTABLE_MAC_CACHE_USE for ARP reply generated by >> the tracked device. >> controller: Use xxreg1 for lookup_nd_ip() and get_nd() actions. >> controller: Send ARP/ND for stale mac_bindings entries. >> > > Hi all, > > We're currently testing this version of the patchset and we're seeing a > some traffic disruption that might or might not be introduced by this > patchset (an OpenShift CI upgrade test failed with this series applied). > > I'd like to request postponing applying this patchset until we get to > the bottom of this. >
After a closer look at the OpenShift CI failure logs and after more re-testing we concluded the single failure we saw is not related to this patch series. From my perspective it's fine to merge this series once code review is completed. Thank you for the patience! Regards, Dumitru _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
