Bug 1522700

Summary: [Extras] Update DPDK to 17.11
Product: Red Hat Enterprise Linux 7 Reporter: Timothy Redaelli <tredaelli>
Component: dpdkAssignee: Timothy Redaelli <tredaelli>
Status: CLOSED ERRATA QA Contact: Jean-Tsung Hsiao <jhsiao>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 7.5CC: atragler, ctrautma, fleitner, hewang, kzhang, narendra_k, ovs-team, pezhang, prabhakar_pujeri, tli, tredaelli, yselkowi
Target Milestone: rcKeywords: Extras, Rebase
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: dpdk-17.11-7.el7 Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2018-04-10 23:59:23 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 1335825, 1518884    
Bug Blocks: 1413149, 1449793, 1461182, 1490967, 1500889    

Description Timothy Redaelli 2017-12-06 09:45:13 UTC

Comment 7 Jean-Tsung Hsiao 2017-12-21 18:15:30 UTC
The package has been tested and passed with the following tests:

# PvP 64 bytes zero loss testing between testpmd and Xena as traffic generator --- vhostuser is installed with DPDK-17.11-4.

# P2P 64 bytes zero loss testing between testpmd/host and Trex --- the host is loaded with DPDK-17.11-4.

# Netperf testing between name spaces on the test driver using testpmd/SUT as loopback.

Comment 8 Marcelo Ricardo Leitner 2018-02-16 15:58:54 UTC
*** Bug 1455140 has been marked as a duplicate of this bug. ***

Comment 9 Marcelo Ricardo Leitner 2018-02-16 16:04:39 UTC
*** Bug 1517210 has been marked as a duplicate of this bug. ***

Comment 10 Flavio Leitner 2018-02-16 16:07:08 UTC
*** Bug 1497384 has been marked as a duplicate of this bug. ***

Comment 11 Christian Trautman 2018-03-01 01:38:47 UTC
Tested http://download-node-02.eng.bos.redhat.com/brewroot/packages/dpdk/17.11/7.el7/x86_64/dpdk-17.11-7.el7.x86_64.rpm

Ran sr-iov testing and standard pvp testing with ovs-dpdk 2.9 from FDP.

Ran testing against 16.11-2 versus 17.11 and found no degradation with Intel x520 cards for 64 or 1500 bytes performance.

Comment 13 Jean-Tsung Hsiao 2018-03-01 12:06:05 UTC
The package has been tested and passed with the following tests:

1. PvP 64 bytes zero loss testing between testpmd and Xena over ixgbe using 2Q/4PMD --- 9.86 Mpps.

2. PvP 64 bytes zero loss testing between testpmd and Trex over 40Gb i40e(XL710) using 2Q/4PMD --- 5.90 Mpps.

Related packages:

Host is using OVS 2.9.0-1 fdP, and guest is using DPDK 17.11-7.

Both host and guest are running under kernel-851.

Comment 15 Jean-Tsung Hsiao 2018-03-02 03:03:58 UTC
(In reply to Jean-Tsung Hsiao from comment #13)
> The package has been tested and passed with the following tests:
> 
> 1. PvP 64 bytes zero loss testing between testpmd and Xena over ixgbe using
> 2Q/4PMD --- 9.86 Mpps.
For 1Q/2PMD the Mpps rate is 5.02

> 
> 2. PvP 64 bytes zero loss testing between testpmd and Trex over 40Gb
> i40e(XL710) using 2Q/4PMD --- 5.90 Mpps.
> 

Also, ran P2P 64 bytes zero loss between testpmd/host and Trex over 40Gb XL710 --- 36.10 Mpps.


> Related packages:
> 
> Host is using OVS 2.9.0-1 fdP, and guest is using DPDK 17.11-7.
> 
> Both host and guest are running under kernel-851.

Comment 16 Pei Zhang 2018-03-02 03:54:00 UTC
Update:

From Virt QE side, all testings with dpdk has finished and get PASS results.

Versions:
kernel-3.10.0-855.el7.x86_64
qemu-kvm-rhev-2.10.0-21.el7.x86_64
tuned-2.9.0-1.el7.noarch
libvirt-3.9.0-13.el7.x86_64
dpdk-17.11-7.el7.x86_64
microcode-20180108.tgz

Network cards: 10-Gigabit X540-AT2

Intel meltdonw&spectre fixes were applied to both the host&guest.
microcode: revision 0x3b, date = 2017-11-17

Values of related options:
# cat /sys/kernel/debug/x86/pti_enabled
1
# cat /sys/kernel/debug/x86/ibpb_enabled
1
# cat /sys/kernel/debug/x86/ibrs_enabled
0
# cat /sys/kernel/debug/x86/retp_enabled
1

Testing Scenarios:
(1) PVP performance testing -- PASS
(Note: dpdk's testpmd acts as the role of OpenvSwitch in host)

The throughput results looks good as expected:
1_Queue/0_Loss/64Byte_packet throughput: 9.49Mpps
1_Queue/0.002%_Loss/64Byte_packet throughput: 17.06Mpps

(2) PVP live migration testing -- PASS

All 10 ping-pong migration works well as expected:
No Stream_Rate Downtime Totaltime Ping_Loss trex_Loss
 0       1Mpps      119     13254        15    9352010.0
 1       1Mpps      123     13453        15    9546293.0
 2       1Mpps      131     12844        15    7015119.0
 3       1Mpps      119     12575        14    4898332.0
 4       1Mpps      124     13021        15    4759253.0
 5       1Mpps      125     13461        16    8348222.0
 6       1Mpps      122     12638        14    6433116.0
 7       1Mpps      121     12581        14    5951345.0
 8       1Mpps      128     13078        15    5130945.0
 9       1Mpps      119     13181        13    6856561.0


(3) Guest with device assignment -- PASS

The throughput results looks good as expected:
1_Queue/0_Loss/64Byte_packet throughput: 20.64Mpps

(4) Guest with ovs+dpdk+vhost-user -- PASS

Other versions:
openvswitch-2.9.0-3.el7fdp.x86_64

The throughput results looks good as expected:
2_Queues/0_Loss/64Byte_packet throughput: 21.30Mpps

Comment 21 errata-xmlrpc 2018-04-10 23:59:23 UTC
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHEA-2018:1065