openvswitch agent ofctl request errors: 'timed out' and 'Datapath Invalid'

Bug #1821753 reported by Oleg Bondarev
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
neutron
New
Undecided
Unassigned

Bug Description

Release: Queens, ovsdb_interface=native, of_request_timeout = 30

With number of OVS ports growing on the node following errors start to occur (starting at ~1200 ports):

ERROR neutron.plugins.ml2.drivers.openvswitch.agent.openflow.native.ofswitch [req-db47426c-1719-43dd-8ecf-4fb4bdcbc316 - - - - -] ofctl request version=None,msg_type=None,msg_len=None,xid=None,OFPFlowMod(buffer_id=4294967295,command=0,cookie=5881109557449606263L,cookie_mask=0,flags=0,hard_timeout=0,idle_timeout=0,instructions=[OFPInstructionActions(actions=[OFPActionPopVlan(len=8,type=18), OFPActionSetField(tunnel_id=725), OFPActionOutput(len=16,max_len=0,port=1793,type=0), OFPActionOutput(len=16,max_len=0,port=2,type=0)],type=4)],match=OFPMatch(oxm_fields={'vlan_vid': 4175}),out_group=0,out_port=0,priority=1,table_id=22) error Datapath Invalid 64183592930369: InvalidDatapath: Datapath Invalid
 or
ERROR neutron.plugins.ml2.drivers.openvswitch.agent.openflow.native.ofswitch [req-632b8ede-1234-4682-afe0-3aefb615b121 - - - - -] ofctl request version=0x4,msg_type=0xe,msg_len=0x78,xid=0x73c67c07,OFPFlow
Mod(buffer_id=4294967295,command=0,cookie=5881109557449606263L,cookie_mask=0,flags=0,hard_timeout=0,idle_timeout=0,instructions=[OFPInstructionActions(actions=[OFPActionPopVlan(len=8,type=18), OFPActionSetField(tunnel_id=666), OFPActionOu
tput(len=16,max_len=0,port=2,type=0)],len=48,type=4)],match=OFPMatch(oxm_fields={'eth_dst': 'fa:16:3e:4a:79:ce', 'vlan_vid': 6107}),out_group=0,out_port=0,priority=2,table_id=20) timed out: Timeout: 30 seconds

with corresponding errors is ovs-vswitchd logs:

|rconn|ERR|br-tun<->tcp:127.0.0.1:6633: no response to inactivity probe after 5 seconds, disconnecting
|rconn|ERR|br-floating<->tcp:127.0.0.1:6633: no response to inactivity probe after 5 seconds, disconnecting
|rconn|ERR|br-int<->tcp:127.0.0.1:6633: no response to inactivity probe after 5 seconds, disconnecting

Setting inactivity_probe to a greater value helps:

#ovs-vsctl set controller br-int inactivity_probe=30000
#ovs-vsctl set controller br-tun inactivity_probe=30000
#ovs-vsctl set controller br-floating inactivity_probe=30000

Should neutron allow setting inactivity_probe for controllers?
Should it correspond to of_request_timeout value?

Revision history for this message
Antonio Ojea (aojea) wrote :
Revision history for this message
Oleg Bondarev (obondarev) wrote :

Right, thanks

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.