Thanks Fabrice, this helps alot.

Ali

On Fri, Oct 26, 2018 at 9:21 AM Durand fabrice <[email protected]> wrote:

> Hello Ali,
>
> reevaluate access will use CoA (wiredeauthtechnique) and restart switch
> will use snmp.
>
> It's how it works in PacketFence.
>
> Regards
>
> Fabrice
>
>
> Le 18-10-24 à 02 h 01, Amjad Ali a écrit :
>
> Hi Fabrice
>
> Thanks for that clarification but what about bouncing port with CoA
> method? When I hit the Restart Switchport button, I want PF to send a CoA
> bounce host port command. Currently as you said, its using the SNMP down
> and then SNMP up to restart switch port. Maybe we need something similar to
> wiredauthTechniques for bounce port such as wiredbounceportTechnique? So as
> to chose between which method to use while bouncing a port. Slightly
> unclear to me on how to go about this please help.
>
> Thanks,
> Ali
>
> On Tue, Oct 23, 2018 at 9:13 PM Fabrice Durand <[email protected]> wrote:
>
>> Hello Ali,
>>
>> in fact /usr/local/pf/html/pfappserver/lib/pfappserver/Model/Node.pm
>> bouncePort is made to shut/no shut the port and it use snmp.
>>
>> What you will need to do is to implement the function
>> wiredeauthTechniques (for wire) or deauthTechniques (for wireless) in order
>> to launch the correct function to reevaluate the access.
>>
>> Regards
>>
>> Fabrice
>>
>>
>> Le 18-10-21 à 23 h 36, Amjad Ali a écrit :
>>
>> Hi Fabrice,
>>
>> Yes your spot on, the issue was wrong port numbers, we'll fix that very
>> soon.
>>
>> A slightly different issue but I need your advice on it, I have changed
>> bouncePort sub routine in Node.pm to send the mac address instead of switch
>> port index for CoA to work properly. This has been done
>> at  /usr/local/pf/html/pfappserver/lib/pfappserver/Model/Node.pm
>>
>>
>> unless($switch->bouncePort($locationlog->{port})) { # changed port to mac
>>         $status = $STATUS::INTERNAL_SERVER_ERROR;
>>         $status_msg = "Couldn't restart port.";
>>     }
>>
>>
>> Need to know what would be the best way to change this preferred behavior
>> from SNMP to CoA. Because later on if we submit this module to be part of
>> PF I guess there would be some issues about it.
>>
>> Many thanks again.
>> Ali
>>
>> On Sat, Oct 20, 2018 at 11:18 AM Durand fabrice via PacketFence-users <
>> [email protected]> wrote:
>>
>>> Hello Ali,
>>>
>>> you did the good thing but in the capture it looks that the switch reply
>>> on the wrong port.
>>>
>>> CoA request: src port 52492 dst 3799
>>>
>>> CoA-ACK : src port 1812 dst 3799 (it's suppose to be src 3799 dst 52492)
>>>
>>> So it looks to me a switch bug.
>>>
>>> Regards
>>>
>>> Fabrice
>>> Le 18-10-15 à 05 h 47, Amjad Ali via PacketFence-users a écrit :
>>>
>>> Hi All,
>>>
>>> We have implemented CoA method to bounce port (reuse Cisco.pm
>>> _radiusBounceMac) in our new hardware module but with some issues.
>>>
>>> 1) The bounce port CoA request packet is received at switch, the switch
>>> replies with CoA-ACK and obliges with port down then port port up. (It does
>>> what its supposed to do, no problems)
>>> 2) The CoA-ACK reply packet also arrives at the switch (I confirmed it
>>> with tcpdump) but packetfence somehow can't get the reply packet. Instead I
>>> get the following log entries
>>>
>>> Oct 15 16:43:59 packetfence httpd_admin: httpd.admin(826) INFO:
>>> [mac:unknown] boucing MAC e0:db:55:cd:84:62 using RADIUS CoA-Request method
>>> (pf::Switch::Pica::bouncePort)
>>> Oct 15 16:43:59 packetfence httpd_admin: httpd.admin(826) WARN:
>>> [mac:unknown] Unable to perform RADIUS CoA-Request: Timeout waiting for a
>>> reply from 10.10.51.217 on port 3799 at /usr/local/pf/lib/pf/util/
>>> radius.pm line 162. (pf::Switch::Pica::catch {...} )
>>> Oct 15 16:43:59 packetfence httpd_admin: httpd.admin(826) ERROR:
>>> [mac:unknown] Cannot restart switch port for e0:db:55:cd:84:62
>>> (pfappserver::PacketFence::Controller::Node::restart_switchport)
>>>
>>> I checked the Radius.pm code (perform_dynauth), it sends the CoA request
>>> packet and listens for a reply, the reply arrives at the machine running
>>> packetfence but evades the socket listening for reply.
>>>
>>> I double checked the timeout and port number but couldn't get to the
>>> root cause. Any ideas would be highly appreciated. I'm attaching the
>>> capture request/reply pcap for your reference. Please advise.
>>>
>>> Thanks,
>>> Ali
>>> --
>>> Amjad Ali
>>>
>>>
>>> _______________________________________________
>>> PacketFence-users mailing 
>>> [email protected]https://lists.sourceforge.net/lists/listinfo/packetfence-users
>>>
>>> _______________________________________________
>>> PacketFence-users mailing list
>>> [email protected]
>>> https://lists.sourceforge.net/lists/listinfo/packetfence-users
>>>
>>
>>
>> --
>> Amjad Ali
>>
>> --
>> Fabrice [email protected] ::  +1.514.447.4918 (x135) ::  
>> www.inverse.ca
>> Inverse inc. :: Leaders behind SOGo (http://www.sogo.nu) and PacketFence 
>> (http://packetfence.org)
>>
>>
>
> --
> Amjad Ali
>
>

-- 
Amjad Ali
_______________________________________________
PacketFence-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/packetfence-users

Reply via email to