Also,
If i restart the ovs and try running the testpmd again,
I see this, from inside the container.
EAL: Detected 4 lcore(s)
EAL: Detected 1 NUMA nodes
EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
EAL: Probing VFIO support...
vhost_user_read(): Failed to recv msg hdr: -1 instead of 12.
vhost_user_sock(): Received msg failed: Connection reset by peer
virtio_user_dev_init(): get_features failed: Connection reset by peer
virtio_user_pmd_probe(): virtio_user_dev_init fails
vdev_probe(): failed to initialize virtio_user1 device
EAL: Bus (vdev) probe failed.
Seems like the daemon is crashing,
And i see this in the ovs-vswitchd.log when the testpmd fails,
2019-11-07T07:31:43.860Z|00001|dpdk|INFO|VHOST_CONFIG: new vhost user
connection is 311
2019-11-07T07:31:43.860Z|00002|dpdk|INFO|VHOST_CONFIG: new device, handle
is 0
2019-11-07T07:31:43.860Z|00003|dpdk|INFO|VHOST_CONFIG: read message
VHOST_USER_SET_OWNER
2019-11-07T07:31:43.860Z|00004|dpdk|INFO|VHOST_CONFIG: read message
VHOST_USER_GET_FEATURES
2019-11-07T07:31:46.265Z|00002|daemon_unix(monitor)|ERR|1 crashes: pid
12315 died, killed (Segmentation fault), core dumped, restarting
and ovs goes for a restart.
Suddenly ovs is not able to read the vhost-user ports when it is coming up.
Here's what 'ovs-vsctl show' returns.
Bridge "br0"
Port "vhost-user4"
Interface "vhost-user4"
type: dpdkvhostuser
error: "could not open network device vhost-user4 (Unknown
error -1)"
Port "br0"
Interface "br0"
type: internal
Port "vhost-user1"
Interface "vhost-user1"
type: dpdkvhostuser
error: "could not open network device vhost-user1 (Unknown
error -1)"
Port "vhost-user3"
Interface "vhost-user3"
type: dpdkvhostuser
error: "could not open network device vhost-user3 (Unknown
error -1)"
Port "vhost-user2"
Interface "vhost-user2"
type: dpdkvhostuser
error: "could not open network device vhost-user2 (Unknown
error -1)"
ovs_version: "2.11.1"
vhost-user ports are back to normal after manually restarting the ovs, but
i still cannot run the testpmd from inside the container and i face the
same issue all over again.
Is this happening because i'm on a pseudo host(I think otherwise)? or did i
miss some steps along the way?
Thanks,
-Deepak Gowda
On Wed, 6 Nov 2019 at 16:47, Deepak Gowda <deepak.d...@gmail.com> wrote:
> Hi,
>
> I'm trying to run test-pmd with this virtual devices created by the
> ovs-dpdk from inside a docker container.
> I see EAL probing failing every time, i did try restarting the ovs, no
> luck.
> I see these vhost-user ports created from the log, i do see these from the
> 'ovs-vsctl show'.
> Can't seem to figure out whats going wrong here,
>
> EAL: Detected 4 lcore(s)
> EAL: Detected 1 NUMA nodes
> EAL: Multi-process socket /var/run/dpdk/rte/mp_socket
> EAL: Probing VFIO support...
> vhost_user_setup(): connect error, Connection refused
> virtio_user_dev_init(): backend set up fails
> virtio_user_pmd_probe(): virtio_user_dev_init fails
> vdev_probe(): failed to initialize net_virtio_user3 device
> vhost_user_setup(): connect error, Connection refused
> virtio_user_dev_init(): backend set up fails
> virtio_user_pmd_probe(): virtio_user_dev_init fails
> vdev_probe(): failed to initialize net_virtio_user4 device
> EAL: Bus (vdev) probe failed.
> testpmd: No probed ethernet devices
>
>
> DPDK ver: 18.11.3
> OVS ver: 2.11.1
> Host OS: debian (although this is not a actual host(bare metal), this is a
> vm really)
>
> Any advise is appreciated.
> Thanks!
> -Deepak Gowda
>
>
>
