On Feb 10, 2014, at 3:28 PM, Jesse Gross wrote:
> On Mon, Feb 10, 2014 at 11:15 AM, Pravin Shelar wrote:
>> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme
>> wrote:
>>> @@ -80,96 +76,126 @@ void ovs_flow_stats_update(struct sw_flow *flow,
>>> struct sk_buff *skb)
>>>tcp_flag
On Feb 10, 2014, at 3:21 PM, Jesse Gross wrote:
> On Mon, Feb 10, 2014 at 11:31 AM, Jarno Rajahalme
> wrote:
>> On Feb 8, 2014, at 7:01 PM, Jesse Gross wrote:
>>> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme
>>> wrote:
diff --git a/datapath/flow.h b/datapath/flow.h
index eafcfd
On Mon, Feb 10, 2014 at 11:15 AM, Pravin Shelar wrote:
> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
>> @@ -80,96 +76,126 @@ void ovs_flow_stats_update(struct sw_flow *flow, struct
>> sk_buff *skb)
>> tcp_flags = TCP_FLAGS_BE16(tcp_hdr(skb));
>> }
>>
>> -
On Mon, Feb 10, 2014 at 11:31 AM, Jarno Rajahalme wrote:
> On Feb 8, 2014, at 7:01 PM, Jesse Gross wrote:
>> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme
>> wrote:
>>> diff --git a/datapath/flow.h b/datapath/flow.h
>>> index eafcfd8..f6cce35 100644
>>> --- a/datapath/flow.h
>>> +++ b/datapat
On Feb 8, 2014, at 7:01 PM, Jesse Gross wrote:
> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
>> diff --git a/datapath/flow.c b/datapath/flow.c
>> index abe6789..e86034e 100644
>> --- a/datapath/flow.c
>> +++ b/datapath/flow.c
>> @@ -80,96 +76,126 @@ void ovs_flow_stats_update(struct
On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
> Keep kernel flow stats for each NUMA node rather than each (logical)
> CPU. This avoids using the per-CPU allocator and removes most of the
>
> diff --git a/datapath/flow.c b/datapath/flow.c
> index abe6789..e86034e 100644
> --- a/d
On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
> diff --git a/datapath/flow.c b/datapath/flow.c
> index abe6789..e86034e 100644
> --- a/datapath/flow.c
> +++ b/datapath/flow.c
> @@ -80,96 +76,126 @@ void ovs_flow_stats_update(struct sw_flow *flow, struct
> sk_buff *skb)
>
> On Feb 6, 2014, at 6:36 PM, Jesse Gross wrote:
>
>> On Thu, Feb 6, 2014 at 4:09 PM, Pravin Shelar wrote:
>>> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme
>>> wrote:
>>>Keep kernel flow stats for each NUMA node rather than each (logical)
>>>CPU. This avoids using the per-CPU allo
On Thu, Feb 6, 2014 at 4:09 PM, Pravin Shelar wrote:
> On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
>> Keep kernel flow stats for each NUMA node rather than each (logical)
>> CPU. This avoids using the per-CPU allocator and removes most of the
>> kernel-side OVS locking ove
On Thu, Feb 6, 2014 at 3:13 PM, Jarno Rajahalme wrote:
> Keep kernel flow stats for each NUMA node rather than each (logical)
> CPU. This avoids using the per-CPU allocator and removes most of the
> kernel-side OVS locking overhead otherwise on the top of perf reports
> and allows
Keep kernel flow stats for each NUMA node rather than each (logical)
CPU. This avoids using the per-CPU allocator and removes most of the
kernel-side OVS locking overhead otherwise on the top of perf reports
and allows OVS to scale better with higher number of threads.
With 9
11 matches
Mail list logo
