Hi Mario, Can you check the offload settings of the tap device that is connected to guest?
Run “ethtool -k tap-solaripub”. On the guest. Raise the log verbosity by going to device manager -> NetKVM device -> Advanced tab -> Logging.Level and changing it to 4. Use DebugView to record the driver tracing (enable kernel trace): http://technet.microsoft.com/en-us/sysinternals/bb896647.aspx Best regards, Yan. On Nov 22, 2013, at 7:28 PM, Mario De Chenno <c...@dechenno.eu> wrote: > Hi all. > We are facing some network issues on some Windows Server 2003 machines. > In short, network seems to lock up and stops responding even to ping > requests. From TCPdump on the tap interface on the server I only see arp > request to the gateway, without replies. Increased tx overruns on the some > interface too. No errors at all on windows side. Most of the time we can > disable the interface from within windows and re-enable it to make things > work again. Rarely we have to shut down the virtual machine (Windows reboot > does not solve the problem). > The issue appears randomly without apparent relation with server activity. > One server makes a lot of small outbond connections and locks up about every > 12-24 hours. It has two nic defined and only one is affected (the heaviest > loaded). Other servers have just one nic and lock less often. > We run Qemu-kvm 1.4.0 and latest virtio-win drivers. On the same host we did > run a Linux web server with no issue in months of activity, with traffic > peaks up to 80Mb/s. > > Here's is the startup command: > > qemu-system-x86_64 -enable-kvm -machine type=pc,accel=kvm -cpu kvm64 -pidfile > /vmstore/vm_pids/solari.pid -rtc base=localtime -drive > file=/vmstore/vm_disks/solari.img,if=virtio -netdev > tap,id=nic01,ifname=tap-solaripub,script=pub102-ifup -device > virtio-net-pci,netdev=nic01,mac=CE:DA:01:00:17:16 -netdev > tap,id=nic02,ifname=tap-solariconsip,script=consip-ifup -device > virtio-net-pci,netdev=nic02,mac=CE:DA:01:00:16:16 -vnc :16 -m 4096 -boot c -k > it -usbdevice tablet -name solari -daemonize > > We tried also the following options without result > ,vhost=off,vnet_hdr=on > ,ioeventfd=on,event_idx=off > > > How can we troubleshoot the issue? > Thanks >