[rbak-nsp] difference between 11.x and 12.x

[Marcin Kuczera]

On 2013-09-02 11:29, Заикин Станислав wrote:
> 07.08.2013 17:27, Marcin Kuczera пишет:
>>
>>>> As between 11.x and 12.x is just 5 days difference, I don't think that
>>>> this bug had been corrected.
>>>> I rather suspect that that there is a difference in error handling, 
>>>> i.e.
>>>> if EU0 has problem with packet than ignore (drop) and continue instead
>>>> of crash ???
>>>> Or maybe I'am wrong ?
>>>
>>> Hello.
>>>        We are faced with a similar problem.
>>>        We have a lot ippa crash on seos 11.x, using the pppoe + 
>>> enhanced
>>>        nat. (se600 with 10ge-4-port linecards)
>>>        On SEOS 12.1.1.2 was only 1 eppa crash (after 1 month uptime).
>>>        On SEOS 12.1.1.4p1 aaad crashed after 5 days of work.
>>>        Found a company that uses non dhcp clips + enhanced nat, they 
>>> have
>>>        ippa crash began after the nat logging enabling.
>>>        You are using the enhanced nat? Nat logging?
>>>
>>> -- 
>>> Vitaly
>>
>> Well, we have almost no nat. I mean - all subscribers have public IP 
>> addresses
>> and just a little, classical "bind interface" network uses NAT 
>> (office purposes).
>>
>> PPA crashes are nasty as whole lincard reloads, just a XCRP process 
>> crash is not a big deal as it recovers quickly...
>>
>> Regards,
>> Marcin
>>
>>
>>
>>
> Do you use vlan on-demand?
> We have no crashes at 12.1.1.2 with many vlans on-demand (q-in-q) and 
> week's crashes at 11.x.
> But after moving to static vlans we have no crashes at 11.1.2.8 and 
> 11.1.2.7p3.
>

Nope, we have just CLIPS/PPPoE subscribers on regular VLANs (single tag).
However, we did QinQ testing (static vlan, no on-demand) but this config 
was cleared
once problems began.

btw, QinQ in demand is something like 10:100-200 ???

Regards,
Marcin



Regards,
Marcin