Hello, I''m a newbie to Xen so don''t flame... I''ve read on this list that there is a bug in the 3.0.3 version which causes the transfer of large files to make the network stall. This is something we are experiencing here because we''re on Cent)S 5, shipped with Xen 3.0.3-25, and I can''t find any documentation on this bug. What are my options to resolve this issue ?! Thanks for the much appreciated help. Luc _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I''ve got a bit of a problem here with my network. I''m using CentOS 5 as a dom0 with many different domUs. I can reproduce this bug at will, so I guess it''s something serious. I''ve also read some reply on this list which told it is a known bug, but I can''t find any trace of it. I share a file via any protocol on any domU. The file has to be of a big enough size so that the upload takes a while. 1 gig usually does the trick. Then, I logon to another machine (physical machine, that is...), and start retreiving it from the domU instance. After some random number of seconds, the network stalls, the link goes down, and the packets start beeing dropped. Luckily, I can logon locally to the dom0 and confirm that the link is down and that packets are dropped. Briging the peth0 interface down and up again doesn''t fix anything, nor does a physical disconnect/reconnect. I have to reboot the whole machine physically to make it work again. It''s not a DoS problem, because I can confirm that I don''t end up with thousands of TIME_WAIT connections. Now, how do I diagnose / fix the problem ? I''m out of ideas and have been looking around for 3 days now... sadly. Xen : 3.0.3-25 Kernel : 2.6.18-8.1.8 Network : bridged Example domU config : name = "XXXXX" memory = "1500" disk = [ ''phy:/dev/XenGuests0/XXXXX,xvda,w'', ] vif = [ ''mac=00:04:75:4f:XX:XX, bridge=xenbr0'', ] uuid = "c3463377-ba64-caf7-87b1-bc34294274b7" bootloader="/usr/bin/pygrub" vcpus=1 cpus="2" on_reboot = ''restart'' on_crash = ''restart'' _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Boudreau Luc schrieb:> I share a file via any protocol on any domU. The file has to be of a big > enough size so that the upload takes a while. 1 gig usually does the > trick. Then, I logon to another machine (physical machine, that is…), > and start retreiving it from the domU instance. After some random number > of seconds, the network stalls, the link goes down, and the packets > start beeing dropped.Try to disable TCP checksum in every domU for every interface that connects to the bridge. Inside the domU''s: ethtool -K eth0 tx off I dont'' know this for CentOS, but on debian-like systems the perfect place is /etc/network/interfaces . As an example: iface eth0 inet static address 192.168.1.5 netmask 255.255.255.0 gateway 192.168.1.1 post-up ethtool -K eth0 tx off You could try this via console first and watch if your network performs better. Stephan _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Thanks for the tip Stephan, Unfortunately, it didn''t fix the wagon. The network still stalls upon the transfer of large files. Also, packets are beeing dropped from the peth0 RX on my dom0. I tried disabling the RX checksum check but it refused to do so and it wouldn''t be wise anyways. Got any other idea ? -----Message d''origine----- De : Stephan Seitz [mailto:s.seitz@netz-haut.de] Envoyé : 26 septembre 2007 18:27 À : Boudreau Luc Cc : xen-users@lists.xensource.com Objet : Re: [Xen-users] Network stall Boudreau Luc schrieb:> I share a file via any protocol on any domU. The file has to be of a > big enough size so that the upload takes a while. 1 gig usually does > the trick. Then, I logon to another machine (physical machine, that > is...), and start retreiving it from the domU instance. After some > random number of seconds, the network stalls, the link goes down, and > the packets start beeing dropped.Try to disable TCP checksum in every domU for every interface that connects to the bridge. Inside the domU''s: ethtool -K eth0 tx off I dont'' know this for CentOS, but on debian-like systems the perfect place is /etc/network/interfaces . As an example: iface eth0 inet static address 192.168.1.5 netmask 255.255.255.0 gateway 192.168.1.1 post-up ethtool -K eth0 tx off You could try this via console first and watch if your network performs better. Stephan _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
Boudreau Luc wrote:> Thanks for the tip Stephan, > > Unfortunately, it didn''t fix the wagon. The network still stalls upon the transfer of large files. Also, packets are beeing dropped from the peth0 RX on my dom0. I tried disabling the RX checksum check but it refused to do so and it wouldn''t be wise anyways. > > Got any other idea ? > >What NIC are you using? I had network problems on Xen / RHEL4 / tg3 driver / some models (not all) of Broadcom NIC. I''m currently using RHEL5 / updated tg3 driver from Broadcom website, and it works fine. You should probably try updating your network driver first. Regards, Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users
I use the Broadcom tg3 driver. I''ll update it and check again. Thanks. -----Message d''origine----- De : xen-users-bounces@lists.xensource.com [mailto:xen-users-bounces@lists.xensource.com] De la part de Fajar A. Nugraha Envoyé : 27 septembre 2007 22:12 À : xen-users@lists.xensource.com Objet : Re: [Xen-users] Network stall Boudreau Luc wrote:> Thanks for the tip Stephan, > > Unfortunately, it didn''t fix the wagon. The network still stalls upon the transfer of large files. Also, packets are beeing dropped from the peth0 RX on my dom0. I tried disabling the RX checksum check but it refused to do so and it wouldn''t be wise anyways. > > Got any other idea ? > >What NIC are you using? I had network problems on Xen / RHEL4 / tg3 driver / some models (not all) of Broadcom NIC. I''m currently using RHEL5 / updated tg3 driver from Broadcom website, and it works fine. You should probably try updating your network driver first. Regards, Fajar _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users _______________________________________________ Xen-users mailing list Xen-users@lists.xensource.com http://lists.xensource.com/xen-users