Hi, Alfredo.

Thanks for the workaround,
but I can't put high stats refresh code to production due to high cpu load.

Ilya.

2014-09-05 13:03 GMT+04:00 Alfredo Cardigliano <[email protected]>:
> Sorry, I forgot to tell you to use -H
>
> Alfredo
>
> On 05 Sep 2014, at 11:02, Alfredo Cardigliano <[email protected]> wrote:
>
>> Hi Ilya
>> please try zcount in svn, there is a temporary workaround for this.
>>
>> Alfredo
>>
>> On 04 Sep 2014, at 18:21, Ilya Gavrilov <[email protected]> wrote:
>>
>>> Hi, Alfredo
>>>
>>> Thank you for the full answer.
>>> It would be great if you find workaround for this.
>>>
>>> Ilya.
>>>
>>> 2014-09-04 14:52 GMT+04:00 Alfredo Cardigliano <[email protected]>:
>>>> Hi Ilya
>>>> unfortunately this is a limitation of the 82576 chipset:
>>>> since ZC (as DNA) drivers are very close to hw, they read drops from NIC 
>>>> counters,
>>>> on this card the counter is able to accumulate up to 4K.
>>>> Each time we read stats (every second in this case) we reset this counter, 
>>>> thus you see 4k drops per second.
>>>> We will try to find a workaround for this.
>>>>
>>>> Alfredo
>>>>
>>>> On 01 Sep 2014, at 16:58, Ilya Gavrilov <[email protected]> wrote:
>>>>
>>>>> Hi Alfredo,
>>>>>
>>>>> I am using PF_RING 6.0.1, intel gigabit adapters:
>>>>> 03:00.0 Ethernet controller: Intel Corporation 82576 Gigabit Network
>>>>> Connection (rev 01)
>>>>> 03:00.1 Ethernet controller: Intel Corporation 82576 Gigabit Network
>>>>> Connection (rev 01)
>>>>>
>>>>> with driver from drivers/PF_RING_aware/intel/igb/igb-5.2.5-zc/src
>>>>> drivers loaded withs script load_driver.sh from this directory
>>>>>
>>>>> packet rate is about 53kpkt/s
>>>>>
>>>>> 1. with not modified zcount, I get next output
>>>>>
>>>>> sudo ./zcount -i zc:eth4 -c 99
>>>>> .....
>>>>> =========================
>>>>> Absolute Stats: 537'361 pkts (0 drops) - 256'858'558 bytes
>>>>> Actual Stats: 53'731.73 pps (0.00 drops) - 0.21 Gbps
>>>>> =========================
>>>>>
>>>>> sudo ./zcount -i eth4 -c 99
>>>>> ....
>>>>> =========================
>>>>> Absolute Stats: 376'166 pkts (0 drops) - 179'807'348 bytes
>>>>> Actual Stats: 53'731.08 pps (0.00 drops) - 0.21 Gbps
>>>>> =========================
>>>>>
>>>>> 2. with modified zcount ( usleep(1) after pfring_zc_recv_pkt line ):
>>>>>
>>>>> sudo ./zcount_slow -i zc:eth4 -c 99
>>>>> ....
>>>>> =========================
>>>>> Absolute Stats: 112'809 pkts (24'570 drops) - 53'922'702 bytes
>>>>> Actual Stats: 18'808.28 pps (4'094.62 drops) - 0.07 Gbps
>>>>> =========================
>>>>>
>>>>> sudo ./zcount_slow -i eth4 -c 99
>>>>> ....
>>>>> =========================
>>>>> Absolute Stats: 150'821 pkts (262'184 drops) - 72'092'438 bytes
>>>>> Actual Stats: 18'834.86 pps (34'896.90 drops) - 0.07 Gbps
>>>>> =========================
>>>>>
>>>>> So, I get wrong drops when run command sudo ./zcount_slow -i zc:eth4 -c 99
>>>>>
>>>>> Ilya.
>>>>>
>>>>> 2014-09-01 18:29 GMT+04:00 Alfredo Cardigliano <[email protected]>:
>>>>>> Hi Ilya
>>>>>> please provide more details about your configuration in order to 
>>>>>> reproduce the issue:
>>>>>> - driver model/version
>>>>>> - ingress packet rate
>>>>>> - command and some output from zcount
>>>>>>
>>>>>> Alfredo
>>>>>>
>>>>>> On 01 Sep 2014, at 15:14, Ilya Gavrilov <[email protected]> wrote:
>>>>>>
>>>>>>> Hello,
>>>>>>>
>>>>>>> I am using pfring in zc mode and want to get number of lost packets on
>>>>>>> interface.
>>>>>>> It seems pfring_zc_stats doesn't get real number in zc mode.
>>>>>>>
>>>>>>> if i add usleep(1); after getting each packet to zcount.c example 
>>>>>>> application
>>>>>>> i get very small Actual stats pps, but Actual stats drop is very
>>>>>>> small, always about 4000.
>>>>>>>
>>>>>>> In non-zc mode i get real number of lost packets.
>>>>>>>
>>>>>>> How can I get real number of lost packets on interface?
>>>>>>>
>>>>>>> Please, help,
>>>>>>> Ilya.
>>>>>>> _______________________________________________
>>>>>>> Ntop-misc mailing list
>>>>>>> [email protected]
>>>>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>>>>>
>>>>>> _______________________________________________
>>>>>> Ntop-misc mailing list
>>>>>> [email protected]
>>>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>>>> _______________________________________________
>>>>> Ntop-misc mailing list
>>>>> [email protected]
>>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>>>
>>>> _______________________________________________
>>>> Ntop-misc mailing list
>>>> [email protected]
>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>> _______________________________________________
>>> Ntop-misc mailing list
>>> [email protected]
>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>
>
> _______________________________________________
> Ntop-misc mailing list
> [email protected]
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
_______________________________________________
Ntop-misc mailing list
[email protected]
http://listgateway.unipi.it/mailman/listinfo/ntop-misc

Reply via email to