Bug 648572

Summary: virtio GSO makes IPv6 very slow
Product: Red Hat Enterprise Linux 5 Reporter: Dan Yasny <dyasny>
Component: kernelAssignee: Thomas Graf <tgraf>
Status: CLOSED ERRATA QA Contact: Virtualization Bugs <virt-bugs>
Severity: high Docs Contact:
Priority: low    
Version: 5.5.zCC: acathrow, akong, fleitner, jbustos, jolsa, jpirko, mwagner, ndai, nhorman, qcai, rdassen, rkhan, sander, tgraf
Target Milestone: rc   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of:
: 683455 (view as bug list) Environment:
Last Closed: 2011-07-21 10:29:51 UTC Type: ---
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:    
Bug Blocks: 683455    
Attachments:
Description Flags
proposed patch none

Description Dan Yasny 2010-11-01 16:45:40 UTC 
Description of problem:
quoting the customer:

We have network issue's between guests and servers not on the same host (can be other guest on other hosts or normale servers).
We discovered that this is related to IPv6 and GSO in the virtio-net driver.
This is reported in also at:
- http://www.mail-archive.com/kvm@vger.kernel.org/msg28045.html

This one looks also related:
- https://bugzilla.redhat.com/show_bug.cgi?id=490266

For a workaround, turning of GSO (options virtio_net gso=0) on the guests solves the problem for linux guests. We are not sure yet how to do that on windows.
However I am sure this problem is generated on the host so it should be a bug in the kernel on the host.

What we see is that the jumbo/super frames generated by the guests (due to GSO being enabled) are not correctly fragmented by the host at the moment it should leave the physical NIC. Disabling GSO/TSO on the NIC with ethtool on the host does not help.


Version-Release number of selected component (if applicable):

2.6.18-194.el5

How reproducible:
always

Steps to Reproduce:
1. configure a virtio_net VM with IPv6
2. configure a http server with IPv6 in same subnet
3. configure wget to use IPv6, and add the http server to hosts file with it's IPv6 address
4. wget a large file - watch the speeds drop drastically
  
Actual results:
very slow speeds (avg of 1-2mbps on a 1Gbps network)


Expected results:
normal download speeds

Additional info:
Comment 4 Neil Horman 2010-11-10 18:53:50 UTC 
Triage assignment.  If you feel this bug doesn't belong to you, or that it cannot be handled in a timely fashion, please contact me for re-assignment
Comment 8 RHEL Program Management 2011-02-04 10:30:20 UTC 
This request was evaluated by Red Hat Product Management for inclusion in a Red
Hat Enterprise Linux maintenance release.  Product Management has requested
further review of this request by Red Hat Engineering, for potential
inclusion in a Red Hat Enterprise Linux Update release for currently deployed
products.  This request is not yet committed for inclusion in an Update
release.
Comment 12 Thomas Graf 2011-03-09 12:54:24 UTC 
Created attachment 483203 [details]
proposed patch
Comment 14 Jarod Wilson 2011-03-16 18:01:48 UTC 
Patch(es) available in kernel-2.6.18-248.el5
Detailed testing feedback is always welcomed.
Comment 20 Amos Kong 2011-06-06 23:23:53 UTC 
According to comment #19, moving this bug to VERIFIED.
Comment 21 errata-xmlrpc 2011-07-21 10:29:51 UTC 
An advisory has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on therefore solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.

http://rhn.redhat.com/errata/RHSA-2011-1065.html