On 03.02.2016 03:32, Joe Stringer wrote:
> On 29 January 2016 at 04:03, Ilya Maximets wrote:
>> Hi.
>>
>> On 29.01.2016 04:02, Daniele Di Proietto wrote:
>>> Hi guys,
>>>
>>> Sorry for resurrecting this thread, but I think we should reconsider
>>> this patch at this point.
>>>
>>> Let me explain:
On 29 January 2016 at 04:03, Ilya Maximets wrote:
> Hi.
>
> On 29.01.2016 04:02, Daniele Di Proietto wrote:
>> Hi guys,
>>
>> Sorry for resurrecting this thread, but I think we should reconsider
>> this patch at this point.
>>
>> Let me explain: I'm still perfectly happy with the fix we applied,
>
Hi.
On 29.01.2016 04:02, Daniele Di Proietto wrote:
> Hi guys,
>
> Sorry for resurrecting this thread, but I think we should reconsider
> this patch at this point.
>
> Let me explain: I'm still perfectly happy with the fix we applied,
> e4e74c3a2b9a("dpif-netdev: Purge all ukeys when reconfigure
Hi guys,
Sorry for resurrecting this thread, but I think we should reconsider
this patch at this point.
Let me explain: I'm still perfectly happy with the fix we applied,
e4e74c3a2b9a("dpif-netdev: Purge all ukeys when reconfigure pmd."),
it is necessary to properly collect stats and it works per
Just post a proposed fix here,
http://openvswitch.org/pipermail/dev/2015-August/058760.html
Any comments are welcomed,
On Mon, Aug 10, 2015 at 11:12 AM, Alex Wang wrote:
> Hey IIya,
>
> Thx for the reply, Daniele helped me reproduced the issue offline and we
> confirmed that this is not DPDK
Hey IIya,
Thx for the reply, Daniele helped me reproduced the issue offline and we
confirmed that this is not DPDK issue.
However, this is also not a ufid issue either...
Let me explain,
First to clear up, with "packets from the same flow (or stream)", I mean
packets with same src/dst MAC, IP,
Hello, Alex.
I think, that this problem doesn't depend on dpdk. It is generic problem with
ukeys.
I found it while testing ODP-OVS from (Linaro Networking Group) without dpdk.
IMHO, there must be opportunity to install ukeys for same flow by different pmd
threads
anyway, because they are reading
Thx for reporting this, I worry this may not be an ovs issue, but to do with
the dpdk driver... Here is my analysis, please correct me if wrong,
When pmd threads are spawned, each of them will take ownership of port
queues in a round robin fashion... In other words, each port queue will
be read
In multiqueue mode several pmd threads may process one
port, but different queues. Flow doesn't depend on queue.
So, while miss upcall processing, all threads (except first
for that port) will receive error = ENOSPC due to
ukey_install failure. Therefore they will not add the flow
to flow_table an
