[dpdk-users] Running DPDK on Google Cloud Compute Engine (GCE)

[Rex Shang]

I used the tool “DPDK_DIR/usertools/dpdk-setup.sh" to do all of the initialization.  It is quite intuitive to use as the steps are laid out.

Before binding, you need to bring down the link.  Looks to me your network device is “up”.  You should “sudo ip link set DEVICE down” before you bind.

Rex
> On Sep 16, 2017, at 12:01 AM, Learner Study <learner.study at gmail.com> wrote:
> 
> Forgot to show the error I see in my CentOS7 setup:
> 
> [root at localhost tools]# ./dpdk-devbind.py -b ixgbe 02:05.0
> Warning - no supported modules(DPDK driver) are loaded
> Routing table indicates that interface 0000:02:05.0 is active. Not modifying
> 
> Please help! Thanks!
> 
> On Sat, Sep 16, 2017 at 12:00 AM, Learner Study <learner.study at gmail.com> wrote:
>> Hello Rex,
>> 
>> Could you please provide steps u used to configure the interfaces and
>> how u r starting test_pmd?
>> I am having somewhat similar issues on Centos7.
>> 
>> Thanks a lot!
>> 
>> On Fri, Sep 15, 2017 at 7:03 PM, Tan, Jianfeng <jianfeng.tan at intel.com> wrote:
>>> Hi Rex,
>>> 
>>> 
>>> On 9/14/2017 6:53 AM, Rex Shang wrote:
>>>> 
>>>> Hi all
>>>> 
>>>> Just wondering if anyone has successfully have DPDK running on VM inside
>>>> Google Cloud.  I’ve tried similar set up on my Mac Book Pro with Virtual Box
>>>> and had success.  Using similar recipe, I can’t have testpmd pass the
>>>> initialization.  Before I spend too much time, just wondering if I am
>>>> working on mission impossible.  ;)  Your input is greatly appreciated.
>>>> 
>>>> Here is my setup:
>>>> Google Cloud Compute Engine
>>>> 4 vCore with 8 GB of memory
>>>> Ubuntu 16.04 LTS
>>>> 3 NIC with 2 used for DPDK (driver: net_virtio); I am a bit doubtful about
>>>> this NIC but so far the NIC is up with “igb_uio"
>>>> 
>>>> Rex
>>>> 
>>>> ------------
>>>> 
>>>> EAL: Detected 4 lcore(s)
>>>> EAL: No free hugepages reported in hugepages-1048576kB
>>>> EAL: Probing VFIO support...
>>>> EAL: VFIO support initialized
>>>> EAL: PCI device 0000:00:04.0 on NUMA socket -1
>>>> EAL:   Invalid NUMA socket, default to 0
>>>> EAL:   probe driver: 1af4:1000 net_virtio
>>>> EAL: PCI device 0000:00:05.0 on NUMA socket -1
>>>> EAL:   Invalid NUMA socket, default to 0
>>>> EAL:   probe driver: 1af4:1000 net_virtio
>>>> EAL: PCI device 0000:00:06.0 on NUMA socket -1
>>>> EAL:   Invalid NUMA socket, default to 0
>>>> EAL:   probe driver: 1af4:1000 net_virtio
>>>> Interactive-mode selected
>>>> USER1: create a new mbuf pool <mbuf_pool_socket_0>: n=163456, size=2176,
>>>> socket=0
>>>> Configuring Port 0 (socket 0)
>>>> Fail to configure port 0
>>>> EAL: Error - exiting with code: 1
>>>>   Cause: Start ports failed
>>> 
>>> 
>>> When you start testpmd, please make sure the option, --disable-hw-vlan, is
>>> added.
>>> 
>>> Besides, you can apply the below patch to enable the debug mode of virtio to
>>> see what's wrong:
>>> 
>>> diff --git a/config/common_base b/config/common_base
>>> index 5e97a08..5863203 100644
>>> --- a/config/common_base
>>> +++ b/config/common_base
>>> @@ -328,10 +328,10 @@ CONFIG_RTE_LIBRTE_DPAA2_DEBUG_TX_FREE=n
>>> # Compile burst-oriented VIRTIO PMD driver
>>> #
>>> CONFIG_RTE_LIBRTE_VIRTIO_PMD=y
>>> -CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_INIT=n
>>> -CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_RX=n
>>> -CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_TX=n
>>> -CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_DRIVER=n
>>> +CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_INIT=y
>>> +CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_RX=y
>>> +CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_TX=y
>>> +CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_DRIVER=y
>>> CONFIG_RTE_LIBRTE_VIRTIO_DEBUG_DUMP=n
>>> 
>>> #
>>>