Mailing List Archive: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?)

About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?)

Jan 25, 2018, 3:52 AM

Post #1 of 6 (986 views)

Hello,

Regarding the output below, does one need a license to be able to run
10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is
already licensed with "enterprise pro")?

===
[marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
Name SystemName Module MAC BusID
NumaNode Status License
ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0 -1
Up NotFound
ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1 -1
Down NotFound
[marco@nprobe ~]$
[marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
Name: ens2f0
Index: 6
Address: D8:D3:85:A0:12:50
Polling Mode: NAPI
Type: Ethernet
Family: Standard NIC
# Bound Sockets: 1
TX Queues: 1
RX Queues: 1
===
?

Regards,
Marco

2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt>:

> Hi Afredo and anyone "listening",
>
> I would like to rule out if pf_ring is the culprit here...
> What would be the correct way to have nprobe use libpcap (or AF_PACKET??)
> mode of getting packets from the NIC?
> Blacklisting pf_ring module from loading?
>
> Thank you
> Marco
>
>
>
> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org>:
>
>> “Absolute Stats” is the total / average number of packets/bytes
>> “Actual Stats” is the current number of packets/bytes (last second)
>>
>> Alfredo
>>
>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt> wrote:
>>
>> Hello list,
>>
>> Any PFRING wizard that can offer clues on where to start troubleshooting
>> this variance between "Absolut Stats" vs "Actual Stats"...?
>>
>> ?
>> ===
>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>> [sudo] password for marco:
>> Using PF_RING v.7.0.0
>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed:
>> 10000Mb/s]
>> # Device RX channels: 1
>> # Polling threads: 1
>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>> =========================
>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>> =========================
>>
>> =========================
>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89 pkt/sec][2'037.38
>> Mbit/sec]
>> =========================
>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03 Gbps]
>> =========================
>>
>> =========================
>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03 pkt/sec][1'529.35
>> Mbit/sec]
>> =========================
>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02 Gbps]
>> =========================
>> ===
>>
>> Thankx
>> Marco
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>
>>
>>
>

Re: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?) [ In reply to ]

deri at ntop

Jan 25, 2018, 4:05 AM

Post #2 of 6 (986 views)

Permalink

Marco
our licenses don’t have a cap on speed, but they are per device family. So if these devices as 10G you need a 10G license

As you’re a university you can mail education@ntop.org <mailto:education@ntop.org> for free licenses

Regards Luca

> On 25 Jan 2018, at 12:52, Marco Teixeira <marco@scom.uminho.pt> wrote:
>
> Hello,
>
> Regarding the output below, does one need a license to be able to run 10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is already licensed with "enterprise pro")?
>
>
> ===
> [marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
> Name SystemName Module MAC BusID NumaNode Status License
> ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0 -1 Up NotFound
> ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1 -1 Down NotFound
> [marco@nprobe ~]$
> [marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
> Name: ens2f0
> Index: 6
> Address: D8:D3:85:A0:12:50
> Polling Mode: NAPI
> Type: Ethernet
> Family: Standard NIC
> # Bound Sockets: 1
> TX Queues: 1
> RX Queues: 1
> ===
> ?
>
> Regards,
> Marco
>
>
> 2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>>:
> Hi Afredo and anyone "listening",
>
> I would like to rule out if pf_ring is the culprit here...
> What would be the correct way to have nprobe use libpcap (or AF_PACKET??) mode of getting packets from the NIC?
> Blacklisting pf_ring module from loading?
>
> Thank you
> Marco
>
>
>
>
> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org <mailto:cardigliano@ntop.org>>:
> “Absolute Stats” is the total / average number of packets/bytes
> “Actual Stats” is the current number of packets/bytes (last second)
>
> Alfredo
>
>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>
>> Hello list,
>>
>> Any PFRING wizard that can offer clues on where to start troubleshooting this variance between "Absolut Stats" vs "Actual Stats"...?
>>
>> ?===
>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>> [sudo] password for marco:
>> Using PF_RING v.7.0.0
>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed: 10000Mb/s]
>> # Device RX channels: 1
>> # Polling threads: 1
>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>> =========================
>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>> =========================
>>
>> =========================
>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89 pkt/sec][2'037.38 Mbit/sec]
>> =========================
>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03 Gbps]
>> =========================
>>
>> =========================
>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03 pkt/sec][1'529.35 Mbit/sec]
>> =========================
>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02 Gbps]
>> =========================
>> ===
>>
>> Thankx
>> Marco
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc

Re: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?) [ In reply to ]

marco at scom

Jan 25, 2018, 11:29 AM

Post #3 of 6 (986 views)

Permalink

Hi Luca,

I have the details and was going to generate de license fo 10Gb/s... but I
was expecting an option for "Standard NIC" and only see Intel, Myricom and
Napatech/DAG/Fiberblaze...
Wich one should I choose to make for my "Standard NIC" that uses PF_RING in
NAPI mode?
?Or maybe there was some misunderstanding here??

?Regards, and thank you for your advice, and time,
Marco?

2018-01-25 12:05 GMT+00:00 Luca Deri <deri@ntop.org>:

> Marco
> our licenses don’t have a cap on speed, but they are per device family. So
> if these devices as 10G you need a 10G license
>
> As you’re a university you can mail education@ntop.org for free licenses
>
> Regards Luca
>
>
> On 25 Jan 2018, at 12:52, Marco Teixeira <marco@scom.uminho.pt> wrote:
>
> Hello,
>
> Regarding the output below, does one need a license to be able to run
> 10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is
> already licensed with "enterprise pro")?
>
>
> ===
> [marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
> Name SystemName Module MAC BusID
> NumaNode Status License
> ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0
> -1 Up NotFound
> ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1
> -1 Down NotFound
> [marco@nprobe ~]$
> [marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
> Name: ens2f0
> Index: 6
> Address: D8:D3:85:A0:12:50
> Polling Mode: NAPI
> Type: Ethernet
> Family: Standard NIC
> # Bound Sockets: 1
> TX Queues: 1
> RX Queues: 1
> ===
> ?
>
>
> Regards,
> Marco
>
>
>
> 2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt>:
>
>> Hi Afredo and anyone "listening",
>>
>> I would like to rule out if pf_ring is the culprit here...
>> What would be the correct way to have nprobe use libpcap (or AF_PACKET??)
>> mode of getting packets from the NIC?
>> Blacklisting pf_ring module from loading?
>>
>> Thank you
>> Marco
>>
>>
>>
>> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org>:
>>
>>> “Absolute Stats” is the total / average number of packets/bytes
>>> “Actual Stats” is the current number of packets/bytes (last second)
>>>
>>> Alfredo
>>>
>>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt> wrote:
>>>
>>> Hello list,
>>>
>>> Any PFRING wizard that can offer clues on where to start troubleshooting
>>> this variance between "Absolut Stats" vs "Actual Stats"...?
>>>
>>> ?
>>> ===
>>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>>> [sudo] password for marco:
>>> Using PF_RING v.7.0.0
>>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed:
>>> 10000Mb/s]
>>> # Device RX channels: 1
>>> # Polling threads: 1
>>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>>> =========================
>>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>>> =========================
>>>
>>> =========================
>>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89 pkt/sec][2'037.38
>>> Mbit/sec]
>>> =========================
>>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03 Gbps]
>>> =========================
>>>
>>> =========================
>>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03 pkt/sec][1'529.35
>>> Mbit/sec]
>>> =========================
>>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02 Gbps]
>>> =========================
>>> ===
>>>
>>> Thankx
>>> Marco
>>> _______________________________________________
>>> Ntop-misc mailing list
>>> Ntop-misc@listgateway.unipi.it
>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>>
>>>
>>>
>>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>
>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>

Re: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?) [ In reply to ]

deri at ntop

Jan 25, 2018, 11:52 AM

Post #4 of 6 (986 views)

Permalink

> On 25 Jan 2018, at 20:29, Marco Teixeira <marco@scom.uminho.pt> wrote:
>
> Hi Luca,
>
> I have the details and was going to generate de license fo 10Gb/s... but I was expecting an option for "Standard NIC" and only see Intel, Myricom and Napatech/DAG/Fiberblaze...
> Wich one should I choose to make for my "Standard NIC" that uses PF_RING in NAPI mode?
> ?Or maybe there was some misunderstanding here??
Yes there is some disconnection Marco

You only need a license for ZC drivers, otherwise you will use vanilla drivers. So pfcount -i zc:ens2f0 requires a ZC drivers, pfcount -i ens2f0 does not

Please see https://github.com/ntop/PF_RING/wiki <https://github.com/ntop/PF_RING/wiki> for details

Regards Luca
>
> ?Regards, and thank you for your advice, and time,
> Marco?
>
>
>
>
> 2018-01-25 12:05 GMT+00:00 Luca Deri <deri@ntop.org <mailto:deri@ntop.org>>:
> Marco
> our licenses don’t have a cap on speed, but they are per device family. So if these devices as 10G you need a 10G license
>
> As you’re a university you can mail education@ntop.org <mailto:education@ntop.org> for free licenses
>
> Regards Luca
>
>
>> On 25 Jan 2018, at 12:52, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>
>> Hello,
>>
>> Regarding the output below, does one need a license to be able to run 10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is already licensed with "enterprise pro")?
>>
>>
>> ===
>> [marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
>> Name SystemName Module MAC BusID NumaNode Status License
>> ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0 -1 Up NotFound
>> ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1 -1 Down NotFound
>> [marco@nprobe ~]$
>> [marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
>> Name: ens2f0
>> Index: 6
>> Address: D8:D3:85:A0:12:50
>> Polling Mode: NAPI
>> Type: Ethernet
>> Family: Standard NIC
>> # Bound Sockets: 1
>> TX Queues: 1
>> RX Queues: 1
>> ===
>> ?
>>
>> Regards,
>> Marco
>>
>>
>> 2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>>:
>> Hi Afredo and anyone "listening",
>>
>> I would like to rule out if pf_ring is the culprit here...
>> What would be the correct way to have nprobe use libpcap (or AF_PACKET??) mode of getting packets from the NIC?
>> Blacklisting pf_ring module from loading?
>>
>> Thank you
>> Marco
>>
>>
>>
>>
>> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org <mailto:cardigliano@ntop.org>>:
>> “Absolute Stats” is the total / average number of packets/bytes
>> “Actual Stats” is the current number of packets/bytes (last second)
>>
>> Alfredo
>>
>>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>>
>>> Hello list,
>>>
>>> Any PFRING wizard that can offer clues on where to start troubleshooting this variance between "Absolut Stats" vs "Actual Stats"...?
>>>
>>> ?===
>>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>>> [sudo] password for marco:
>>> Using PF_RING v.7.0.0
>>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed: 10000Mb/s]
>>> # Device RX channels: 1
>>> # Polling threads: 1
>>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>>> =========================
>>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>>> =========================
>>>
>>> =========================
>>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89 pkt/sec][2'037.38 Mbit/sec]
>>> =========================
>>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03 Gbps]
>>> =========================
>>>
>>> =========================
>>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03 pkt/sec][1'529.35 Mbit/sec]
>>> =========================
>>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02 Gbps]
>>> =========================
>>> ===
>>>
>>> Thankx
>>> Marco
>>> _______________________________________________
>>> Ntop-misc mailing list
>>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>>
>>
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc

Re: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?) [ In reply to ]

marco at scom

Jan 25, 2018, 12:54 PM

Post #5 of 6 (986 views)

Permalink

Ok. And one can expect to reach more than 1Gb/s on vanilla drivers right?
On a somewhat decent server... Xeon with PCIe x8 NIC...

?Regards
Marco?

2018-01-25 19:52 GMT+00:00 Luca Deri <deri@ntop.org>:

>
>
> On 25 Jan 2018, at 20:29, Marco Teixeira <marco@scom.uminho.pt> wrote:
>
> Hi Luca,
>
> I have the details and was going to generate de license fo 10Gb/s... but I
> was expecting an option for "Standard NIC" and only see Intel, Myricom and
> Napatech/DAG/Fiberblaze...
> Wich one should I choose to make for my "Standard NIC" that uses PF_RING
> in NAPI mode?
> ?Or maybe there was some misunderstanding here??
>
> Yes there is some disconnection Marco
>
> You only need a license for ZC drivers, otherwise you will use vanilla
> drivers. So pfcount -i zc:ens2f0 requires a ZC drivers, pfcount -i ens2f0
> does not
>
> Please see https://github.com/ntop/PF_RING/wiki for details
>
> Regards Luca
>
>
> ?Regards, and thank you for your advice, and time,
> Marco?
>
>
>
> 2018-01-25 12:05 GMT+00:00 Luca Deri <deri@ntop.org>:
>
>> Marco
>> our licenses don’t have a cap on speed, but they are per device family.
>> So if these devices as 10G you need a 10G license
>>
>> As you’re a university you can mail education@ntop.org for free licenses
>>
>> Regards Luca
>>
>>
>> On 25 Jan 2018, at 12:52, Marco Teixeira <marco@scom.uminho.pt> wrote:
>>
>> Hello,
>>
>> Regarding the output below, does one need a license to be able to run
>> 10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is
>> already licensed with "enterprise pro")?
>>
>>
>> ===
>> [marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
>> Name SystemName Module MAC BusID
>> NumaNode Status License
>> ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0
>> -1 Up NotFound
>> ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1
>> -1 Down NotFound
>> [marco@nprobe ~]$
>> [marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
>> Name: ens2f0
>> Index: 6
>> Address: D8:D3:85:A0:12:50
>> Polling Mode: NAPI
>> Type: Ethernet
>> Family: Standard NIC
>> # Bound Sockets: 1
>> TX Queues: 1
>> RX Queues: 1
>> ===
>> ?
>>
>>
>> Regards,
>> Marco
>>
>>
>>
>> 2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt>:
>>
>>> Hi Afredo and anyone "listening",
>>>
>>> I would like to rule out if pf_ring is the culprit here...
>>> What would be the correct way to have nprobe use libpcap (or
>>> AF_PACKET??) mode of getting packets from the NIC?
>>> Blacklisting pf_ring module from loading?
>>>
>>> Thank you
>>> Marco
>>>
>>>
>>>
>>> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org>:
>>>
>>>> “Absolute Stats” is the total / average number of packets/bytes
>>>> “Actual Stats” is the current number of packets/bytes (last second)
>>>>
>>>> Alfredo
>>>>
>>>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt> wrote:
>>>>
>>>> Hello list,
>>>>
>>>> Any PFRING wizard that can offer clues on where to start
>>>> troubleshooting this variance between "Absolut Stats" vs "Actual Stats"...?
>>>>
>>>> ?
>>>> ===
>>>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>>>> [sudo] password for marco:
>>>> Using PF_RING v.7.0.0
>>>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed:
>>>> 10000Mb/s]
>>>> # Device RX channels: 1
>>>> # Polling threads: 1
>>>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>>>> =========================
>>>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>>>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>>>> =========================
>>>>
>>>> =========================
>>>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>>>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89
>>>> pkt/sec][2'037.38 Mbit/sec]
>>>> =========================
>>>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03
>>>> Gbps]
>>>> =========================
>>>>
>>>> =========================
>>>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>>>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03
>>>> pkt/sec][1'529.35 Mbit/sec]
>>>> =========================
>>>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02
>>>> Gbps]
>>>> =========================
>>>> ===
>>>>
>>>> Thankx
>>>> Marco
>>>> _______________________________________________
>>>> Ntop-misc mailing list
>>>> Ntop-misc@listgateway.unipi.it
>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>>>
>>>>
>>>>
>>>
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>
>>
>>
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>
>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc
>

Re: About pf_ring licences (was nProbe Pro won't do more then 1Gb/s?) [ In reply to ]

deri at ntop

Jan 25, 2018, 12:59 PM

Post #6 of 6 (986 views)

Permalink

Marco
I suggest you to start reading the PF_RING articles on our blog as the discussion is pretty long. First of all think in terms of pps and not Gbps, then consider the configuration of nprobe, the nature of traffic (fragments?) etc. Packet capture is just one component

Luca

> On 25 Jan 2018, at 21:54, Marco Teixeira <marco@scom.uminho.pt> wrote:
>
> Ok. And one can expect to reach more than 1Gb/s on vanilla drivers right? On a somewhat decent server... Xeon with PCIe x8 NIC...
>
> ?Regards
> Marco?
>
>
>
> 2018-01-25 19:52 GMT+00:00 Luca Deri <deri@ntop.org <mailto:deri@ntop.org>>:
>
>
>> On 25 Jan 2018, at 20:29, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>
>> Hi Luca,
>>
>> I have the details and was going to generate de license fo 10Gb/s... but I was expecting an option for "Standard NIC" and only see Intel, Myricom and Napatech/DAG/Fiberblaze...
>> Wich one should I choose to make for my "Standard NIC" that uses PF_RING in NAPI mode?
>> ?Or maybe there was some misunderstanding here??
> Yes there is some disconnection Marco
>
> You only need a license for ZC drivers, otherwise you will use vanilla drivers. So pfcount -i zc:ens2f0 requires a ZC drivers, pfcount -i ens2f0 does not
>
> Please see https://github.com/ntop/PF_RING/wiki <https://github.com/ntop/PF_RING/wiki> for details
>
> Regards Luca
>
>>
>> ?Regards, and thank you for your advice, and time,
>> Marco?
>>
>>
>>
>>
>> 2018-01-25 12:05 GMT+00:00 Luca Deri <deri@ntop.org <mailto:deri@ntop.org>>:
>> Marco
>> our licenses don’t have a cap on speed, but they are per device family. So if these devices as 10G you need a 10G license
>>
>> As you’re a university you can mail education@ntop.org <mailto:education@ntop.org> for free licenses
>>
>> Regards Luca
>>
>>
>>> On 25 Jan 2018, at 12:52, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>>
>>> Hello,
>>>
>>> Regarding the output below, does one need a license to be able to run 10Gb/s speeds (please mind, I'm only talking about pf_ring. nprobe is already licensed with "enterprise pro")?
>>>
>>>
>>> ===
>>> [marco@nprobe ~]$ sudo pfcount -L -v1 | grep ens2f
>>> Name SystemName Module MAC BusID NumaNode Status License
>>> ens2f0 ens2f0 pf_ring D8:D3:85:A0:12:50 0000:13:00.0 -1 Up NotFound
>>> ens2f1 ens2f1 pf_ring D8:D3:85:A0:12:54 0000:13:00.1 -1 Down NotFound
>>> [marco@nprobe ~]$
>>> [marco@nprobe ~]$ cat /proc/net/pf_ring/dev/ens2f0/info
>>> Name: ens2f0
>>> Index: 6
>>> Address: D8:D3:85:A0:12:50
>>> Polling Mode: NAPI
>>> Type: Ethernet
>>> Family: Standard NIC
>>> # Bound Sockets: 1
>>> TX Queues: 1
>>> RX Queues: 1
>>> ===
>>> ?
>>>
>>> Regards,
>>> Marco
>>>
>>>
>>> 2018-01-18 10:35 GMT+00:00 Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>>:
>>> Hi Afredo and anyone "listening",
>>>
>>> I would like to rule out if pf_ring is the culprit here...
>>> What would be the correct way to have nprobe use libpcap (or AF_PACKET??) mode of getting packets from the NIC?
>>> Blacklisting pf_ring module from loading?
>>>
>>> Thank you
>>> Marco
>>>
>>>
>>>
>>>
>>> 2018-01-17 22:16 GMT+00:00 Alfredo Cardigliano <cardigliano@ntop.org <mailto:cardigliano@ntop.org>>:
>>> “Absolute Stats” is the total / average number of packets/bytes
>>> “Actual Stats” is the current number of packets/bytes (last second)
>>>
>>> Alfredo
>>>
>>>> On 17 Jan 2018, at 21:13, Marco Teixeira <marco@scom.uminho.pt <mailto:marco@scom.uminho.pt>> wrote:
>>>>
>>>> Hello list,
>>>>
>>>> Any PFRING wizard that can offer clues on where to start troubleshooting this variance between "Absolut Stats" vs "Actual Stats"...?
>>>>
>>>> ?===
>>>> [marco@nprobe ~]$ sudo pfcount -i ens2f0
>>>> [sudo] password for marco:
>>>> Using PF_RING v.7.0.0
>>>> Capturing from ens2f0 [mac: D8:D3:85:A0:12:50][if_index: 5][speed: 10000Mb/s]
>>>> # Device RX channels: 1
>>>> # Polling threads: 1
>>>> Dumping statistics on /proc/net/pf_ring/stats/3096-ens2f0.3
>>>> =========================
>>>> Absolute Stats: [136'980 pkts total][0 pkts dropped][0.0% dropped]
>>>> [136'980 pkts rcvd][126'559'478 bytes rcvd]
>>>> =========================
>>>>
>>>> =========================
>>>> Absolute Stats: [274'202 pkts total][0 pkts dropped][0.0% dropped]
>>>> [274'202 pkts rcvd][254'708'653 bytes rcvd][274'163.89 pkt/sec][2'037.38 Mbit/sec]
>>>> =========================
>>>> Actual Stats: [137'222 pkts rcvd][1'000.13 ms][137'202.92 pps][1.03 Gbps]
>>>> =========================
>>>>
>>>> =========================
>>>> Absolute Stats: [411'199 pkts total][0 pkts dropped][0.0% dropped]
>>>> [411'199 pkts rcvd][382'383'683 bytes rcvd][205'575.03 pkt/sec][1'529.35 Mbit/sec]
>>>> =========================
>>>> Actual Stats: [136'997 pkts rcvd][1'000.09 ms][136'983.43 pps][1.02 Gbps]
>>>> =========================
>>>> ===
>>>>
>>>> Thankx
>>>> Marco
>>>> _______________________________________________
>>>> Ntop-misc mailing list
>>>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>>>
>>>
>>> _______________________________________________
>>> Ntop-misc mailing list
>>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>>
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>>
>> _______________________________________________
>> Ntop-misc mailing list
>> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
>> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it <mailto:Ntop-misc@listgateway.unipi.it>
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc <http://listgateway.unipi.it/mailman/listinfo/ntop-misc>
>
> _______________________________________________
> Ntop-misc mailing list
> Ntop-misc@listgateway.unipi.it
> http://listgateway.unipi.it/mailman/listinfo/ntop-misc