Fuel migration fails on partitioning stage

[Alisa Tselovalnikova]

Detailed bug description:
 Fuel migration fails on partitioning stage with following error:
   + dd if=/dev/vda4 bs=1048576
   + ssh 10.109.0.8 'dd of=/dev/vda4 bs=1048576'
   Warning: Permanently added '10.109.0.8' (ECDSA) to the list of known
            hosts.
   Write failed: Broken pipe
   dd: error writing 'standard output': Broken pipe

Steps to reproduce:
 1. Deploy cluster
 2. Run fuel migration:
        fuel-migration <ip_compute>

Expected results:
 Fuel migrated to a compute.

Actual result:
 Fuel migration fails.
 Full log - http://paste.openstack.org/show/536100/

Reproducibility:
 2 of 3 runs

Description of the environment:
 http://paste.openstack.org/show/536101/

[Alisa Tselovalnikova (atselovalnikova)]

/var/log/messages

[Alisa Tselovalnikova (atselovalnikova)]

/var/log/messages from compute node where Fuel was migrating to

[Peter Zhurba (pzhurba)]

According provided logs. Main issue is "crash of e1000 driver on compute node"
Problem is described in https://bugs.launchpad.net/fuel/+bug/1576276/comments/6
"According 1 there is only common recommendation - update software on Jenkins slave host, increase memory and cpus on compute node"

Also it is possible to change test code for using virtio net instead of e1000 for test environment.