On Tue, Jul 8, 2008 at 5:16 PM, David S. Ahern <[EMAIL PROTECTED]> wrote: > There's a bug opened for the network lockups -- see > http://sourceforge.net/tracker/index.php?func=detail&aid=1802082&group_id=180599&atid=893831 > > Based on my testing I've found that the e1000 has the lowest overhead > (e.g., lowest irq and softirq times in the guest). I have not seen any > lockups with the network using the e1000 nic, and a couple of months ago > I was able to run a reasonably intensive network load continuously for > several days. > > However, the duration tests I've run were with a modified BIOS. Months > ago when I was digging into the network lockups I was comparing > interrupt allocations to a DL320G3 running a RHEL3/4 load natively. I > noticed no interrupts were shared on bare hardware, while in my RHEL3/4 > based kvm guests I was seeing interrupt sharing. So, I patched the bios > (see attached) to get a different usage. > > I have not had time to do the due diligence to see if the stability was > due to kvm updates or my bios change. If you have the time I'd be > interested in knowing how the bios change works for you -- if you still > see lockups.
This bug report is similar to the issue I'm seeing. In our case, I'm booting off a 32-bit Knoppix 5.3 DVD ISO, mounting the virtual partitions, and running rsync from another server on the network. Everything is connected via gigabit NICs and switch ports. Host has a kvmbr0 using bond0 as the physical interface. bond0 combines the 4 ports on an Intel PRO/1000MT PCIe NIC, using mode=balance-tlb. Host is running 64-bit Debian Lenny, with kvm-70 packages and 2.6.24 kernel, using the kvm/kvm-amd modules that ship with the kernel. Hardware: Tyan h2000M motherboard 2x dual-core Opteron 2220 CPUs at 2.8 GHz 8 GB ECC DDR2-667 SD-RAM (4 GB per socket) 12x 500 GB SATA-II HDs in RAID6 3Ware 9650-ML16 PCIe RAID controller The guests are using -net tap. Using rtl8139, I can run rsync until the cows come home (it runs through cron twice a day, but I've done manual runs 6 times back-to-back, to sync 400 GB of data). Using e1000, the guest networking will die within minutes of starting rsync, everytime. Won't last more than 15 minutes. ifdown/ifup eth0 will bring the link back to life, but the rsync process has to be restarted. Using virtio-net (booting the guest OS using kernel 2.6.24, not Knoppix), the guest networking dies within minutes as well, but it lasts a little longer than e1000, and is considerably faster. Guests are started with: /usr/bin/kvm -name webmail -smp 1 -m 3072 -vnc :05 -daemonize -localtime -usb -usbdevice tablet -net nic,macaddr=00:16:3e:00:00:05,model=rtl8139 -net tap,ifname=tap05 -pidfile /var/run/kvm/webmail.pid -boot d -no-reboot -drive index=0,media=disk,if=ide,file=/dev/mapper/vol0-webmail--boot -drive index=1,media=disk,if=ide,file=/dev/mapper/vol0-webmail--storage -drive index=2,media=cdrom,if=ide,file=/home/iso/KNOPPIX_V5.3.1DVD-2008-03-26-EN.iso Number of guests running doesn't make a difference, happens with just one or all 6 running. But only the network for 1 guest dies at a time. -- Freddie Cash [EMAIL PROTECTED] -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
