[c-nsp] ISR detects interface status with delay

[Martin T]

Thanks! Configuring keepalive messages interval to 1 second(default is
10s) indeed decreases polling interval to 1 second. For example if I
configure "no keepalive", then ISR Ethernet interface will never go
down while the cable is physically disconnected. However, why does
decreasing the keepalive messages interval cause this polling interval
to decrease? Is this just some sort of (legacy) code in IOS which
creates this relation between keepalive messages interval and
interface status polling interval? As far as I knew, keepalive
frames(ethertype 0x9000, same src and dst MAC) were just used for
loop-detection in case of older Ethernet cabling standards..


thanks,
Martin

On 2/8/15, Swapnendu Mazumdar <ccie19804 at gmail.com> wrote:
> keepalive 1
>
> under ethernet interface on ISR.
>
> On Sun, Feb 8, 2015 at 4:40 AM, Martin T <m4rtntns at gmail.com> wrote:
>
>> Lukasz,
>>
>> thanks for the reply! Just to make sure, there isn't a way to decrease
>> this polling interval on ISR platform, is there?
>>
>>
>> Martin
>>
>> On Sat, Feb 7, 2015 at 11:42 AM, Łukasz Bromirski <lukasz at bromirski.net>
>> wrote:
>> >
>> >> On 06 Feb 2015, at 10:36, Martin T <m4rtntns at gmail.com> wrote:
>> >>
>> >> In order to illustrate this behavior I made a short video where IOS
>> >> detects that interface Fa0/0 went down with almost 13 seconds delay.
>> >> Usually this delay is around 5 to 8 seconds, but sometimes up to 15
>> >> seconds. Video can be seen here:
>> >> https://www.dropbox.com/s/yb9o379935s3hou/20150206_110347.mp4 PHY chip
>> >> should detect this within micro- or milliseconds and as seen from the
>> >> video, LED's indicating the link status went off immediately when I
>> >> removed the cable, but why does it take so long for IOS to detect
>> >> this?
>> >
>> > First of all, carrier-delay on ISRs is not supported. The command is
>> > for
>> > that hardware platforms, that have more extensive instrumentation
>> > on the edge between hardware and software.
>> >
>> > Second, ISR polls interface controllers for up/down, so your
>> > better bet would be to use BFD, despite it being CPU-intensive,
>> > to detect link up/down event.
>> >
>> > --
>> > "There's no sense in being precise when |               Łukasz
>> > Bromirski
>> >  you don't know what you're talking     |
>> > jid:lbromirski at jabber.org
>> >  about."               John von Neumann |
>> > http://lukasz.bromirski.net
>> >
>>
>> _______________________________________________
>> cisco-nsp mailing list  cisco-nsp at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-nsp
>> archive at http://puck.nether.net/pipermail/cisco-nsp/
>>
>