With RSS i can only have 16 queues (hardware limitation), so I need to use
zbalance_ipc. I'm testing it tomorrow and let you know the results.

Thanks again.

On Tue, Jun 30, 2015 at 2:42 PM, Jose Vila <[email protected]> wrote:

> I've been able to raise the hugepages number to 2048, so I can start all
> the Snort instances I need, but the libnuma "node 9 not allowed" warning
> still persist.
>
> Any ideas or advice on this?
>
> Regards,
>
> Jose Vila.
>
> On Tue, Jun 30, 2015 at 2:02 PM, Jose Vila <[email protected]> wrote:
>
>> Hi Alfredo, and thank you very much for your answer.
>>
>> I've tested the RSS queues option (loaded the ixgbe kernel module with
>> parameter "RSS=22", as I need 22 Snort instances running), but I still have
>> several problems.
>>
>> * On one hand I'm still getting the libnuma "Warning: node 9 not allowed"
>> error. I have no idea on how to solve this.
>> * On the other, I can only start 8 Snort instances. Snort takes all the
>> available hugepages and doesn't let further instances to be run. I've tried
>> configuring a higher number of Hugepages (up to 8192, via sysctl and
>> via /etc/pf_ring/hugepages) but I seem to only have 512 hugepages available
>> (It says so on /proc/meminfo). I'm on CentOS 6.6 with
>> kernel 2.6.32-504.23.4.el6.x86_64.
>>
>> Did I load the ixgbe kernel module the right way? Am I missing something
>> else to get more hugepages?
>>
>> Each of the snort instances is launched with this command (changing
>> interface's RSS queue, bindcpu, clusterid and logdir):
>>
>> /usr/local/snort/bin/snort -c /usr/local/snort/etc/snort.conf -i zc:eth0@1
>> --daq pfring_zc --daq-mode passive --daq-dir /usr/local/lib/daq/ --daq-var
>> bindcpu=1 --daq-var clusterid=1 -R .RED1 -l /var/log/snort/red1 -G 1 -u
>> root -g root -D
>>
>>
>> The output of my startup script:
>>
>> # /etc/init.d/granjero_zc start
>> Lanzando Snorts.
>> Lanzando Snort 1
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>> Spawning daemon child...
>> My daemon child 9415 lives...
>> Daemon parent exiting (0)
>> Lanzando Snort 2
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>> Spawning daemon child...
>> My daemon child 9458 lives...
>> Daemon parent exiting (0)
>> Lanzando Snort 3
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>> Spawning daemon child...
>> My daemon child 9469 lives...
>> Daemon parent exiting (0)
>> [ ... ]
>> Lanzando Snort 9
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>>  *** error retrieving hugepages info ***
>> Lanzando Snort 10
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>>  *** error retrieving hugepages info ***
>> Lanzando Snort 11
>> libnuma: Warning: node 9 not allowed
>> numa_sched_setaffinity_v2_int() failed; abort
>> : Invalid argument
>> set_mempolicy: Invalid argument
>>  *** error retrieving hugepages info ***
>> [ ... ]
>>
>>
>>
>>
>> On Mon, Jun 29, 2015 at 6:48 PM, Alfredo Cardigliano <
>> [email protected]> wrote:
>>
>>> Hi Jose
>>> since ZC is a kernel-bypass technology, which directly access the
>>> network card, only 1 application at a time can access a device/queue.
>>> You have 2 options in order to distribute the load across multiple snort
>>> instances:
>>> 1. load the driver with multiple RSS queues, then start one snort
>>> instance per queue: zc:eth0@0, zc:eth0@1, zc:eth0@2, and so on
>>> 2. load the driver with a single queue, then use zbalance_ipc to
>>> distribute the traffic across multiple software SPSC queues
>>>
>>> Alfredo
>>>
>>>
>>> _______________________________________________
>>> 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