There are many enqueue errors on enabling multiqueue or multiport in VM

Bug #1784525 reported by Yi-y-yang
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Juniper Openstack
Status tracked in Trunk
Trunk
In Progress
High
Yi-y-yang
OpenContrail
New
Undecided
Unassigned

Bug Description

Here is "vif --get 4 --rate" output when I run pktgen with 3 ports and 2 queues per port.

vif0/5 PMD: tapb8293d4d-d6
            Type:Virtual HWaddr:00:00:5e:00:01:00 IPaddr:1.1.1.4
            Vrf:4 Flags:PL3L2D QOS:-1 Ref:23
            RX device packets:0 bytes:0 errors:0
            RX port packets:9181090 errors:0
            RX queue packets:2517180 errors:6663956
            RX queue errors to lcore 0 0 0 0 0 0 0 0 0 0 823863 1030365 920639 345863 598704 1278289 857358 808872
            RX packets:2517099 bytes:151025994 errors:0
            TX packets:0 bytes:0 errors:0
            Drops:683358192
            TX queue packets:0 errors:0
            TX port packets:0 errors:0
            TX device packets:0 bytes:0 errors:0

The reason is forwarding thread just dequeues once and consumes those packets one forwarding thread enqueued every loop, but the actual case other forwarding threads also enqueue some packets, so gradually lcore rx ring will have more and more packets accumulating there, so enqueue errors will be inevitable.

Ideally, every forwarding thread should consume all the packets enqueued by other forwarding threads once.

Tags: vrouter
Revision history for this message
OpenContrail Admin (ci-admin-f) wrote : [Review update] master

Review in progress for https://review.opencontrail.org/45097
Submitter: Yi Yang (<email address hidden>)

Jeba Paulaiyan (jebap)
tags: added: vrouter
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.