hi, we have some problems regarding to networking on xen-unstable. we are using xen since 2.0.5 now and its really great. due the fine experience the last months we are buying bigger servers now :) now we have a problem with a dual xeon/6gb machine. we can get 6gb working with the pae extension or networking running (on 2.0.7) but not pae and networking on xen-unstable (source from yesterday and tested the last days). the mailinglist archiv shows much problems on the network side, but no fixes for this (our) problem. box is: 2*2.8xeon (fsb 800) with intel chipset 7230, 6x1gb ram, single e100 and dual e1000 (only the first e1000 active) and 150gb raid5 on icp/gdth with debian 3.1 (sarge) as base system. dom0 and domU boots fine except networking in domU does not work. on 2.0.7 everything works correct except the 6gb memory allocation. no errors are reported. routing etc. looks fine. i was looking for some information of the new peth0 interface but did not found anything usefull, does it break the setup? attached you can find the network setup, kernel config and the xend.log. as u can see, xen-br0 does recieve packets but not send anything. do you need more information? thanks for your help and keep up the really good work, xen is great an makes my admin day much easier :) -- Tschaw/2, Stoeni! ++ Please remember: rm -rf means "read mail -really fast" ++ _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 15 Oct 2005, at 22:13, stoeni wrote:> i was looking for some information of the new peth0 interface but did > not found anything usefull, does it break the setup? > > attached you can find the network setup, kernel config and the > xend.log. > > as u can see, xen-br0 does recieve packets but not send anything. > > do you need more information?The transmit statistic on xen-br0 is a red herring: looking at my working system, the xen-br0 interface also reports 0 packets transmitted. Odd, but obviously therefore not an indication of trouble. It sounds like dom0 networking is okay for you. In cases where our bridging setup doesn''t work, I believe it is usual for networking to fail completely (both dom0 and domU). When you boot a domU, do you have a vif called something like ''vif1.1'' in dom0''s ifconfig listing? Does this interface show any packets sent or received? Does ''brctl show'' show vif1.1 attached to xen-br0? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
keir, Keir Fraser schrieb:> The transmit statistic on xen-br0 is a red herring: looking at my > working system, the xen-br0 interface also reports 0 packets > transmitted. Odd, but obviously therefore not an indication of trouble.ok, good to know.> It sounds like dom0 networking is okay for you. In cases where our > bridging setup doesn''t work, I believe it is usual for networking to > fail completely (both dom0 and domU).dom0 networking is fine. i see a zombie process in the processtable: USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 2188 0.0 0.0 0 0 ? Z Oct15 0:00 [network-bridge] <defunct>> When you boot a domU, do you have a vif called something like ''vif1.1'' > in dom0''s ifconfig listing? Does this interface show any packets sent or > received?yes, recieved and transmit packets: vif0.0 Link encap:Ethernet HWaddr FE:FF:FF:FF:FF:FF UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:1708751 errors:0 dropped:0 overruns:0 frame:0 TX packets:913713 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2540344455 (2.3 GiB) TX bytes:64017546 (61.0 MiB) peth0 also has traffic in and out.> Does ''brctl show'' show vif1.1 attached to xen-br0?yes, it does: root@xen300 [tmp] # brctl show bridge name bridge id STP enabled interfaces xen-br0 8000.feffffffffff no peth0 vif0.0 looking on a 2.0.7 bridge it looks different: root@xen207 [tmp] # brctl show bridge name bridge id STP enabled interfaces xen-br0 8000.00e018526ab6 no eth0 vif2.0 maybe problems with this peth0 thing? should the bridge be attachted to eth0 instead of peth0? here is some dmesg output from dom0: e1000: eth0: e1000_watchdog_task: NIC Link is Up 100 Mbps Full Duplex device vif0.0 entered promiscuous mode bridge: can''t decode speed from peth0: 65535 device peth0 entered promiscuous mode xen-br0: port 1(vif0.0) entering learning state xen-br0: topology change detected, propagating xen-br0: port 1(vif0.0) entering forwarding state e1000: peth0: e1000_watchdog_task: NIC Link is Up 100 Mbps Full Duplex xen-br0: port 2(peth0) entering learning state xen-br0: topology change detected, propagating xen-br0: port 2(peth0) entering forwarding state the 100mbps link is ok, this machine currently runs in a test environment. anything else i should look at to track this problem down? -- Tschaw/2, Stoeni! ++ Please remember: rm -rf means "read mail -really fast" ++ _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 16 Oct 2005, at 10:11, stoeni wrote:>> When you boot a domU, do you have a vif called something like ''vif1.1'' >> in dom0''s ifconfig listing? Does this interface show any packets sent >> or >> received? > > yes, recieved and transmit packets: > > vif0.0 Link encap:Ethernet HWaddr FE:FF:FF:FF:FF:FFNo, that''s part of domain0''s network. You are looking for something like vifx.1 (e.g., vif1.1, vif2.1, etc.). If you don''t have one of those, your domain has been set up with no network. Which would probably be due to a problem in the config file. -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
On 16 Oct 2005, at 10:36, stoeni wrote:>> No, that''s part of domain0''s network. > > hm, this is new to xen 3.0? > > on 2.0.7 there are only vif1.x interfaces, no vif0.0. did i miss > something in the docs?Yes, peth0 and vif0.0 are new in 3.0.>> You are looking for something like vifx.1 (e.g., vif1.1, vif2.1, >> etc.). > > no interface like this, just startet the domU again. > > only eth0, peth0, vif0.0 and xen-br0.That''s the problem, then.>> If you don''t have one of those, your domain has been set up with no >> network. Which would probably be due to a problem in the config file. > > config used from 2.0.7:Hmmm... it looks okay to me. Maybe you can stop xend, delete /var/log/xen*.log, the start xend try booting your domain, and see what errors you get in /var/log/xend.log and /var/log/xend-debug.log? -- Keir _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser schrieb:> On 16 Oct 2005, at 10:11, stoeni wrote: >> vif0.0 Link encap:Ethernet HWaddr FE:FF:FF:FF:FF:FF > No, that''s part of domain0''s network.hm, this is new to xen 3.0? on 2.0.7 there are only vif1.x interfaces, no vif0.0. did i miss something in the docs?> You are looking for something like vifx.1 (e.g., vif1.1, vif2.1, etc.).no interface like this, just startet the domU again. only eth0, peth0, vif0.0 and xen-br0.> If you don''t have one of those, your domain has been set up with no > network. Which would probably be due to a problem in the config file.config used from 2.0.7: #=============================kernel = "/boot/vmlinuz-2.6-xenU" root = "/dev/sda1 ro" memory = 2048 name = "vm-tempo" cpu = -1 # leave to Xen to pick nics = 1 vif = [ ''bridge=xen-br0'' ] disk [''phy:/dev/vg_xlarge/tempo,sda1,w'',''phy:/dev/vg_xlarge/tempo_swap,sda2,w''] restart = ''onreboot'' #============================= -- Tschaw/2, Stoeni! ++ Please remember: rm -rf means "read mail -really fast" ++ _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel
Keir Fraser schrieb:> On 16 Oct 2005, at 15:27, stoeni wrote: >> how can we track down the problem now, was it "silent" fixed by the new >> version or did i something wrong at compiling? > > > Maybe you had a slightly inconsistent kernel image hanging around. If > the bug is still around it will show up again and we''ll go after it > then. For now, let''s be glad it''s gone! :-)i did some compile stuff again and again on the source from last night. i''m not really sure but it looks like disabling smp does it. maybe some other option for the kernels (networking stuff). maybe its the hardware, brandnew intel board with xeons on fsb800 ... (intel premium partner) crazy little thing called xen 3.0 ... ;-) i''ll try some compile options the next days (if i get some time to spend on it) and report the results here. many thanks for your help! -- Tschaw/2, Stoeni! Fog Lamps, n.: Excessively (often obnoxiously) bright lamps mounted on the fronts of automobiles; used on dry, clear nights to indicate that the driver''s brain is in a fog. See also "Idiot Lights". _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel