This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-users] massive packet loss under network and disk load

To: xen-users@xxxxxxxxxxxxxxxxxxx
Subject: Re: [Xen-users] massive packet loss under network and disk load
From: "Ryan Burke" <burke@xxxxxxxxxxxxxxxxx>
Date: Wed, 23 Apr 2008 07:18:23 -0500 (CDT)
Delivery-date: Wed, 23 Apr 2008 05:19:01 -0700
Envelope-to: www-data@xxxxxxxxxxxxxxxxxx
Importance: Normal
In-reply-to: <480E9FC3.1020208@xxxxxxxxx>
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/cgi-bin/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
References: <480E9FC3.1020208@xxxxxxxxx>
Reply-to: burke@xxxxxxxxxxxxxxxxx
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: SquirrelMail/1.4.13
> As mentioned on a post to a thread about drbd and Xen I've been seeing
> some horrific packet loss under network and disk IO load which has been
> causing huge problems for drbd syncing and replication.
> After testing many parameters and settings I have found that this
> appears to make the problem go away:
> vif = [
>          'rate=20MB/s,mac=AA:01:06:6D:FD:02, bridge=xenbr2'
>   ]
> 'this' being the 'rate=20MB/s'.
> I'm not sure if this is merely brushing the problem under the carpet
> but, all other things being at default settings, adding and removing the
> rate option from the vif reliably makes the problem go away or come back.
> The thing that worries me is that even with this parameter on the vif I
> still see what I'd call massive packet loss on the vif for this domU as
> seen in xm top.
> All the dropped packets are RX
> It looks like this:
> Net0 RX: 544045974bytes  6253667pkts        0err    52645drop  TX:
> 334397858bytes  3644125pkts
> I hope that I am interpreting this correctly, that the '52645drop'
> indicates RX packets dropped...
> I am seeing hundreds of drops per second.
> Within the domU, ifconfig shows no dropped packets or errors.
> On dom0, ifconfig shows TX packets dropped:
> vif2.0    Link encap:Ethernet  HWaddr FE:FF:FF:FF:FF:FF
>            inet6 addr: fe80::fcff:ffff:feff:ffff/64 Scope:Link
>            UP BROADCAST RUNNING NOARP  MTU:1500  Metric:1
>            RX packets:4187429 errors:0 dropped:0 overruns:0 frame:0
>            TX packets:7189734 errors:0 dropped:60679 overruns:0 carrier:0
>            collisions:0 txqueuelen:0
>            RX bytes:383941238 (366.1 MiB)  TX bytes:1911795684 (1.7 GiB)

Just out of curiosity have you disabled TX checksums? I didn't follow your
previous thread so I don't know that others had suggested. Try "ethtool -
K eth0 tx off" and see if that helps your TX packet loss.


Xen-users mailing list

<Prev in Thread] Current Thread [Next in Thread>