[ kvm-Bugs-1816417 ] Networking fails after migration (two or more virtual NICs)

[ kvm-Bugs-1816417 ] Networking fails after migration (two or more virtual NICs)

[SourceForge.net]

Bugs item #1816417, was opened at 2007-10-19 10:33
Message generated for change (Tracker Item Submitted) made by Item Submitter
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=1816417&group_id=180599

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: None
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Richard Fuller (rp_fuller)
Assigned to: Nobody/Anonymous (nobody)
Summary: Networking fails after migration (two or more virtual NICs)

Initial Comment:
Using kvm-48 with the module from kvm-48, on a 2.6.23.1 kernel. Same problem if I use a 2.6.22.9 kernel, or if I use the kvm module from either kernel. Using a 64-bit host and a 32-bit guest.

With one virtual NIC live migration works fine. With two, after the live migration, packets intended for the second interface go out on the first, and packets intended for the first interface disappear. With three virtual NICs, if only the first two are brought up the same behaviour is experienced. If the third is brought up the live migration fails. (It appears to succeed, but the KVM process on the target is hung.)

I am starting KVM like so:

kvm -net nic,macaddr=02:00:90:20:B0:1F,vlan=176 -net tap,script=/nfs/vms/ifup.176,vlan=176 \
    -net nic,macaddr=02:00:C0:A8:6E:83,vlan=310 -net tap,script=/nfs/vms/ifup.310,vlan=310 \
    -kernel /nfs/vms/kernel \
    -hda /tmp/image \
    -initrd /nfs/vms/tftp.initrd \
    -append "root=/dev/ram0 ramdisk_size=32768 console=ttyS0" \
    -m 128

The ifup scripts just do (with 176 replaced as appropriate for 310):
/sbin/ifconfig $1 0.0.0.0 promisc up
/sbin/brctl addif br176 $1

The first is brought up as 144.32.176.31, the second as 192.168.110.131.

If I send some pings, and monitor tap0/tap1 on the physical host, I get:

# tcpdump -i tap0 icmp

18:29:28.462710 IP 144.32.176.31 > 144.32.176.1: ICMP echo request, id 48387, seq 1, length 64
18:29:28.463460 IP 144.32.176.1 > 144.32.176.31: ICMP echo reply, id 48387, seq 1, length 64

# tcpdump -i tap1 icmp

18:29:13.539600 IP 192.168.110.131 > 192.168.110.32: ICMP echo request, id 47875, seq 6, length 64
18:29:13.539936 IP 192.168.110.32 > 192.168.110.131: ICMP echo reply, id 47875, seq 6, length 64

all working fine.

I start a kvm on the target, same parameters, but with:

-incoming tcp://0:4444

as well, then do a migration from the source with:

migrate tcp://192.168.110.205:4444

With two NICs the migration succeeds, and I can use the virtual serial console on the target. However, the networking no longer functions. Packets intended for the second interface go out on the first, and packets intended for the first interface disappear...

# tcpdump -i tap0 icmp

18:24:30.580316 IP 192.168.110.131 > 192.168.110.32: ICMP echo request, id 48643, seq 6, length 64

# tcpdump -i tap1 icmp

(silence)

Am I doing something wrong, or is there a bug?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=893831&aid=1816417&group_id=180599

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/