[Users] Bonding - VMs Network performance problem
Ricardo Esteves
maverick.pt at gmail.com
Tue Jul 30 16:32:09 UTC 2013
Hi,
Just found the problem / solution.
http://www.novell.com/support/kb/doc.php?id=7007658
https://bugzilla.redhat.com/show_bug.cgi?id=518531
It's the LRO activated on bnx2x driver.
After the reboot the LRO comes activated, after reconfiguring the bond
the LRO is deactivated.
So i created /etc/modprobe.d/bnx2x.conf with: options bnx2x
disable_tpa=1
and everythink is ok now after reboot.
Best regards,
Ricardo Esteves.
-----Original Message-----
From: Ricardo Esteves <maverick.pt at gmail.com>
To: Users at ovirt.org
Cc: Itamar Heim <iheim at redhat.com>, Mike Kolesnik <mkolesni at redhat.com>
Subject: Re: [Users] Bonding - VMs Network performance problem
Date: Tue, 30 Jul 2013 16:58:44 +0100
Hi,
I also noticed this on /var/log/messages while the problem was happening
(i was downloading a file in the VM from the web):
Jul 30 16:55:55 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:56 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:56 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:56 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:56 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:56 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:57 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:57 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:57 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Jul 30 16:55:57 blade5 kernel: bond0.1: received packets cannot be
forwarded while LRO is enabled
Best regards,
Ricardo Esteves.
-----Original Message-----
From: Ricardo Esteves <maverick.pt at gmail.com>
To: Users at ovirt.org
Cc: Itamar Heim <iheim at redhat.com>, Mike Kolesnik <mkolesni at redhat.com>
Subject: Re: [Users] Bonding - VMs Network performance problem
Date: Tue, 30 Jul 2013 15:10:24 +0100
Good afternoon,
In attachment the result of vdsClient -s 0 getVdsCaps during problem and
after problem resolved.
Best regards,
Ricardo Esteves.
-----Original Message-----
From: Mike Kolesnik <mkolesni at redhat.com>
To: Ricardo Esteves <maverick.pt at gmail.com>
Cc: Users at ovirt.org, Itamar Heim <iheim at redhat.com>
Subject: Re: [Users] Bonding - VMs Network performance problem
Date: Tue, 16 Jul 2013 10:01:31 -0400 (EDT)
----- Original Message -----
> On 07/16/2013 04:09 PM, Ricardo Esteves wrote:
> > Hi,
> >
> > Not really.
> >
> > I can resolve temporally, unconfiguring the bond and then configure it
> > again.
> >
> > But when i reboot the server the problem comes back.
>
> can you compare the network configuration before and after you change it
> the setup network?
Perhaps you can send pastebin of a 'vdsClient 0 getVdsCaps' (or 'vdsClient -s 0 getVdsCaps' if the first doesn't work)
from the host when the bond is slow, and one after you break and create it again?
if you don't have vdsClient command you can 'yum install vdsm-cli' and it should be available.
>
> >
> >
> > -----Original Message-----
> > *From*: Itamar Heim <iheim at redhat.com
> > <mailto:Itamar%20Heim%20%3ciheim at redhat.com%3e>>
> > *To*: Ricardo Esteves <maverick.pt at gmail.com
> > <mailto:Ricardo%20Esteves%20%3cmaverick.pt at gmail.com%3e>>
> > *Subject*: Re: [Users] Bonding - VMs Network performance problem
> > *Date*: Tue, 16 Jul 2013 15:38:10 +0300
> >
> > On 07/01/2013 03:12 PM, Ricardo Esteves wrote:
> >> Hi,
> >>
> >> Yes, i'm still experiencing this problem, in fact just happened a few
> >> minutes ago. :)
> >>
> >> All MTUs are 1500.
> >
> > was this resolved?
> >
> >>
> >> -----Original Message-----
> >> *From*: Livnat Peer <lpeer at redhat.com <mailto:lpeer at redhat.com>
> >> <mailto:Livnat%20Peer%20%3clpeer at redhat.com%3e>>
> >> *To*: Ricardo Esteves <maverick.pt at gmail.com
> >> <mailto:maverick.pt at gmail.com>
> >> <mailto:Ricardo%20Esteves%20%3cmaverick.pt at gmail.com%3e>>
> >> *Subject*: Re: [Users] Bonding - VMs Network performance problem
> >> *Date*: Sun, 23 Jun 2013 11:33:58 +0300
> >>
> >> Hi Ricardo,
> >> Are you still experiencing the problem described below?
> >> Are you configuring MTU (to something other than default or 1500) for
> >> one of the networks on the bond?
> >>
> >> Thanks, Livnat
> >>
> >> On 06/18/2013 05:36 PM, Ricardo Esteves wrote:
> >>> Good afternoon,
> >>>
> >>> Yes, the "Save network configuration" is checked, configurations are
> >>> persistent across boots.
> >>>
> >>> The problem is not the persistence of the configurations, the problem is
> >>> that after a reboot the network performance on the VMs is very bad, and
> >>> to fix it i need to remove the bonding and add it again.
> >>>
> >>> In attachment, the screenshots of my network configuration.
> >>>
> >>> Best regards,
> >>> Ricardo Esteves.
> >>>
> >>> -----Original Message-----
> >>> *From*: Mike Kolesnik <mkolesni at redhat.com <mailto:mkolesni at redhat.com>
> >>> <mailto:mkolesni at redhat.com>
> >>> <mailto:Mike%20Kolesnik%20%3cmkolesni at redhat.com%3e>>
> >>> *To*: Ricardo Esteves <maverick.pt at gmail.com
> >>> <mailto:maverick.pt at gmail.com> <mailto:maverick.pt at gmail.com>
> >>> <mailto:Ricardo%20Esteves%20%3cmaverick.pt at gmail.com%3e>>
> >>> *Cc*:Users at ovirt.org <mailto:Users at ovirt.org> <mailto:Users at ovirt.org>
> >>> <mailto:Users at ovirt.org>
> >>> *Subject*: Re: [Users] Bonding - VMs Network performance problem
> >>> *Date*: Sun, 26 May 2013 04:57:43 -0400 (EDT)
> >>>
> >>>
> >>>
> >>> ------------------------------------------------------------------------
> >>>
> >>>
> >>> Hi,
> >>>
> >>> I've got ovirt installed on 2 HP BL460c G6 blades, and my VMs have
> >>> very poor network performance (around 7,01K/s).
> >>>
> >>> On the servers itselfs there is no problem, i can download a file
> >>> with wget at around 99 M/s.
> >>>
> >>> Then i go to ovirt network configuration remove the bonding and then
> >>> make the bonding again and the problem gets fixed (i have to do this
> >>> everytime i reboot my blades).
> >>>
> >>> Have you tried to check the "Save network configuration" check box, or
> >>> clicking the button from the host's NICs sub-tab?
> >>>
> >>> This should persist the configuration that you set on the host across
> >>> reboots..
> >>>
> >>>
> >>> SERVER' s Software:
> >>> CentOS 6.4 (64 bits) - 2.6.32-358.6.2.el6.x86_64
> >>> Ovirt EL6 official rpms.
> >>>
> >>> Anyone experienced this kind of problems?
> >>>
> >>> Best regards,
> >>> Ricardo Esteves.
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Users mailing list
> >>>Users at ovirt.org <mailto:Users at ovirt.org> <mailto:Users at ovirt.org>
> >>>http://lists.ovirt.org/mailman/listinfo/users
> >>>
> >>>
> >>>
> >>>
> >>>
> >>> _______________________________________________
> >>> Users mailing list
> >>>Users at ovirt.org <mailto:Users at ovirt.org> <mailto:Users at ovirt.org>
> >>>http://lists.ovirt.org/mailman/listinfo/users
> >>>
> >>
> >>
> >>
> >> _______________________________________________
> >> Users mailing list
> >>Users at ovirt.org <mailto:Users at ovirt.org>
> >>http://lists.ovirt.org/mailman/listinfo/users
> >>
> >
>
> _______________________________________________
> Users mailing list
> Users at ovirt.org
> http://lists.ovirt.org/mailman/listinfo/users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/users/attachments/20130730/9f172484/attachment-0001.html>
More information about the Users
mailing list