Description of problem:
I do tunnelled migration with 1024 round and since about 500 round it will report an error
# virsh migrate --live --p2p --tunnelled kvm-rhel6-i386 qemu+ssh://10.66.85.231/system
error: internal error Cannot extract Qemu version from '/usr/libexec/qemu-kvm'

Version-Release number of selected component (if applicable):
# rpm -qa libvirt qemu-kvm kernel
kernel-2.6.32-131.0.5.el6.x86_64
qemu-kvm-0.12.1.2-2.161.el6.x86_64
libvirt-0.8.7-18.el6.x86_64

How reproducible:
always

Steps to Reproduce:
1. Prepare 2 hosts and prepare a nfs which is mounted on both hosts, and setting the virt_use_nfs boolean on both sides
# setsebool -P virt_use_nfs 1
and close the iptable on both sides
# iptables -F

2. start a domain on source host using nfs
3. sh migrate.sh <guestname> <source_host_ip> <target_host_ip>
cat migrate.sh
#!/bin/bash

GUEST=$1
HOST1=$2
HOST2=$3
OPTIONS="--live --p2p --tunnelled"
TRANSPORT="ssh"

date
for i in `seq 1 1024`;
do
    echo "Loop ${i}: Migrating ${GUEST} from ${HOST1} to ${HOST2}"
    echo "COMMAND: virsh migrate ${OPTIONS} ${GUEST} qemu+${TRANSPORT}://${HOST2}/system"
    time virsh migrate ${OPTIONS} ${GUEST} qemu+${TRANSPORT}://${HOST2}/system >> /tmp/mig_result 2>&1 >> /tmp/mig_result

    echo "Loop ${i}: Migrating ${GUEST} back from ${HOST2} to ${HOST1}"
    echo "COMMAND: virsh -c qemu+${TRANSPORT}://${HOST2}/system migrate ${OPTIONS} ${GUEST} qemu+${TRANSPORT}://${HOST1}/system"
    time virsh -c qemu+${TRANSPORT}://${HOST2}/system migrate ${OPTIONS} ${GUEST} qemu+${TRANSPORT}://${HOST1}/system 2>&1 >> /tmp/mig_result
done
date
  
4. see screen output and /tmp/mig_result file

Actual results:
in the end, there will be errors when doing migration from target to source
error: internal error Cannot extract Qemu version from '/usr/libexec/qemu-kvm'

Expected results:
all 1024 round migration will succeed with no error

Additional info: