[ovs-discuss] ovs-vswitch.log miss_upcall and ARP reply issue

[Leigh Woolley]

Hi All,



Just wondering where to start looking for a answer to the following issue:



When performing an ARP request on a specific host (11.0.0.1) we're seeing the responses returned on the expected virtual NIC via tcpdump. However the ARP request isn't been passed up to the controller. 


The switch and controller are running inside a qemu VM running under GNS3.  The 11.0.0.1 host VM is running an FRR. Other containers connected to the switch seem to behave as expected.



Enabling debugging in the vswitchd logs we're seeing the following entry every time an arp request is made.



019-02-27T04:01:14.457Z|00070|dpif(handler7)|DBG|system at ovs-system: miss upcall:
recirc_id(0),dp_hash(0),skb_priority(0),in_port(3),skb_mark(0),ct_state(0),ct_zone(0),ct_mark(0),ct_label(0),eth(src=0c:78:0a:9a:65:00,dst=0c:20:35:97:6b:00),eth_type(0x0806),arp(sip=11.0.0.1,tip=11.0.0.2,op=2,sha=0c:78:0a:9a:65:00,tha=0c:20:35:97:6b:00)
arp,vlan_tci=0x0000,dl_src=0c:78:0a:9a:65:00,dl_dst=0c:20:35:97:6b:00,arp_spa=11.0.0.1,arp_tpa=11.0.0.2,arp_op=2,arp_sha=0c:78:0a:9a:65:00,arp_tha=0c:20:35:97:6b:00


I'm guessing this is the issue but what could cause the upcall to be missed?



We're  running ovs 2.8.1 under Fedora 28. 



Thanks

Leigh
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.openvswitch.org/pipermail/ovs-discuss/attachments/20190227/85ecb2cc/attachment.html>