you might be able to filter by port, like you do with by ip, but I've never
tried it
On Thu, Feb 25, 2016 at 12:32 AM, Todd Millecam <
tyggna@gmail.com> wrote:
> Also, tc doesn't support port-specific operations as that's all handled in
> a different space of the kernel. Only iptables can do port-specific
> traffic shaping. You'd have to do something really clever like direct all
> non port 22 traffic from specified ip to a tap interface off of a bridge
> and then use tc on that interface, but if you can't mess with iptables
> without lots of paperwork, I imagine making virtual network adapters public
> is also not a possibility.
>
> On Thu, Feb 25, 2016 at 12:26 AM, Todd Millecam <tyggna@gmail.com> wrote:
>
>> well, for that, then it's a:
>> match ip src/dst 10.208.208.0/21
>> appended onto your tc.
>> To emulate loss, you'd use netem as well and just append a loss 1% to the
>> end.
>>
>> Lastly, to get a proper range you'd want to change delay 100ms to the
>> average and then the range, so like:
>> delay 175ms 75ms loss 1%
>>
>> It's all in the same tc module, so man tc-netem will be a lot more
>> helpful than me.
>>
>> On Wed, Feb 24, 2016 at 6:17 PM, Bryan O'Neal <
>> Bryan.ONeal@theonealandassociates.com> wrote:
>>
>>> Can't mess with IP tables on the server. Well I could but that would
>>> require a week of paperwork ;)
>>>
>>> Something like match ip src 10.208.208.0/21 match ip should work in tc
>>> but how do I say not port? I know I can say sport but not sure about port
>>> and I have no idea how to say "not port"
>>>
>>> as for latency range delay takes 2 arguments so it would be netem delay
>>> 100ms 150ms would be from 100ms-250ms delay. IIRC
>>>
>>> Also you are using outbound/root, which I know is more full featured.
>>> Trying to get something to work on inbound... I think I may just be too
>>> tired and should probably call it a day and try tomorrow... Unless an
>>> expert show up with a magic pill for me so I don't have to think at 7am :)
>>>
>>> On Wed, Feb 24, 2016 at 6:06 PM, Todd Millecam <tyggna@gmail.com> wrote:
>>>
>>>> tc qdisc add dev eth0 root netem delay 100ms
>>>> iptables -A INPUT -m statistic --mode random --probability 0.01 -j DROP
>>>>
>>>>
>>>> That should get you started
>>>>
>>>>
>>>> On Wed, Feb 24, 2016 at 5:52 PM, Bryan O'Neal <
>>>> Bryan.ONeal@theonealandassociates.com> wrote:
>>>>
>>>>> I am looking for a tc command that will add 100-250ms of latency
>>>>> to all traffic that is not on port 22
>>>>> to/from an ip range like 10.208.208.0/21.
>>>>>
>>>>> Bonus:
>>>>> I would also like 0-1% packet loss
>>>>> I would like this on inbound (ingress) not outbound (root) traffic
>>>>>
>>>>> Any one able to help?
>>>>>
>>>>> ---------------------------------------------------
>>>>> PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
>>>>> To subscribe, unsubscribe, or to change your mail settings:
>>>>> http://lists.phxlinux.org/mailman/listinfo/plug-discuss
>>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> Todd Millecam
>>>>
>>>> ---------------------------------------------------
>>>> PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
>>>> To subscribe, unsubscribe, or to change your mail settings:
>>>> http://lists.phxlinux.org/mailman/listinfo/plug-discuss
>>>>
>>>
>>>
>>> ---------------------------------------------------
>>> PLUG-discuss mailing list - PLUG-discuss@lists.phxlinux.org
>>> To subscribe, unsubscribe, or to change your mail settings:
>>> http://lists.phxlinux.org/mailman/listinfo/plug-discuss
>>>
>>
>>
>>
>> --
>> Todd Millecam
>>
>
>
>
> --
> Todd Millecam
>
--
Todd Millecam
---------------------------------------------------
PLUG-discuss mailing list -
PLUG-discuss@lists.phxlinux.org
To subscribe, unsubscribe, or to change your mail settings:
http://lists.phxlinux.org/mailman/listinfo/plug-discuss
(text/plain)