From: Anthony Sequeira (terry.francona@gmail.com)
Date: Fri Dec 02 2005 - 20:24:06 GMT-3
Thank heavens for this group!!!!!!
Once again I am pulled back from the edge thanks to GroupStudy! When I get
my CCIE I am donating like $43,567 to the guy that runs this thing.
Thanks so much Andy....I now see the penalty is indeed 1000 - but that it
basically starts decaying right away due to the exponential nature of the
decay.
Thanks for clarifying my knowledge of this feature, for pointing out the
error in Cisco's documentation (which appears to be limited to their stating
of the default half-life), and for helping me regain my faith in this
journey!
On 12/2/05, Edwards, Andrew M <andrew.m.edwards@boeing.com> wrote:
>
> Interesting... tha'ts not what I'm seeing.
>
> I'm looking at this:
>
> int f0/0
> dampen 15
> shut
>
> As soon as I shut the interface down the penalty goes to 1000. Then
> begins to decay at a half life of 15 seconds. The IOS may seem that its
> decaying faster, but its doing it correctly. I say the IOS looks faster
> because the decay is actually an exponentially decreasing logarithmic decay
> function based on a half life of 15 seconds. So you should see the penalty
> begin to decrement immediately according to the exponential decay function.
>
> That's all fine and dandy, but I think test wise, its important to just
> reason out what is being asked of you.
>
> For example, if you are asked to dampen the interface when it flaps 3
> times in 30 seconds, you should realize with a default half life of 5
> seconds, you will never make it dampen becuase the penalty decays too fast.
> I think a more appropriate answer would be "dampen 30". This way the
> compound penalty will be some number >2000 within 30 seconds.
>
> Here is a long winded explaination of how I know the compound penalty is >
> 2000. First flap at t=0 (penalty is 1000). For argument sake, say the
> interface flaps again at t+15seconds. Without doing the full math, it
> stands to reason that the accumulated penalty is greater than 500 (because
> we haven't reached the half life of 30 seconds yet) but less than 1000. So
> the new penalty is greater than 1500 but less than 2000.
>
> And now lets go one step further. For argument sake, lets say that the
> interface flaps one more time at say t+29 seconds. Again, we haven't
> reached the original half life, nor have we reached the second penalty half
> life. And for this reason, the first half life penalty cannot be less than
> 500 (because this would be the half life) and neither can the second half
> life penalty be less than 500.
>
> So, I have 2 penalties that are greater than 500. The summation of these
> two penalties will yield a penalty that is greater than 1000. And with the
> final penalty on the third flap of 1000, you can see we will be at a
> compound penalty somewhere greater than 2000.
>
> Since 2000 is the penalty marker to dampen the interface, we were
> successful at meeting the requirement of dampening the interface if it
flaps
> 3 times within 30 seconds.
>
> I tried to find a good site that discussed exponential decay, but its been
> a while since I looked at logarithmic functions so they made no sense to
> me... but knowing the general rule of half life you can get a range as I
> dicsussed above that should help guide the course.
>
> HTH,
>
> andy
>
> -----Original Message-----
> *From:* Anthony Sequeira [mailto:terry.francona@gmail.com]
> *Sent:* Friday, December 02, 2005 1:59 PM
> *To:* Edwards, Andrew M
> *Cc:* my-ccie-test@libero.it; ccielab@groupstudy.com
> *Subject:* Re: IP Event Dampening
>
> Yes - I saw that the half life default is actually 5 seconds.
>
> But I also noticed that the penalty did not appear to be 1000.
>
> Also - I set the half life to 15 seconds - yet the penalty decay had
> nothing to do with 15 seconds. It would decay much more rapidly than that.
>
> It almost looks like Cisco is treating an interface shutdown differently
> than some other interface failure.
>
> How in the world are we to know how Cisco grades this feature???? I can
> only imagine the response I would get from the proctor on this one...."Do
> what the question states please. Thanks."
>
>
> On 12/2/05, Edwards, Andrew M <andrew.m.edwards@boeing.com> wrote:
> >
> > Anthony,
> >
> > Hopefully you saw that the deafult is 5 seconds and not 15 seconds as
> > documented?
> >
> > To verify, use the command:
> >
> > Show interface dampening
> >
> > Seen on 12.2(15)T and even on 12.3
> >
> > HTH,
> >
> > Andy
> > CCIE #15334
> >
> > -----Original Message-----
> > From: Anthony Sequeira [mailto: terry.francona@gmail.com]
> > Sent: Friday, December 02, 2005 12:27 PM
> > To: my-ccie-test@libero.it
> > Cc: ccielab@groupstudy.com
> > Subject: Re: IP Event Dampening
> >
> >
> > Sad news everyone - I just labbed this up as well using 12.2(15)T16
> > code.....NONE of it seems to work as documented.
> >
> > Indeed the default values are not what is documented....AND - it took me
> >
> > like 25 flaps before I could get the interface to dampen. The penalty
> > assigned did not appear to be anything close to 1000 - and the penalty
> > would decay at a rate much faster than the configured or default Half
> > Life.
> >
> > I only pray that Cisco grades this feature based on the documentation -
> > and not on how the feature actually behaves in production.
> >
> > These are the things that make this journey rather frustrating and at
> > times
> > - in fact - bordering on ludicrous.
> >
> >
> >
> > On 12/2/05, my-ccie-test@libero.it <my-ccie-test@libero.it> wrote:
> > >
> > > Hi Anthony,
> > > I did a lab cheking for this feature and wath I saw is the following:
> > >
> > > first of all doing the command show interface dampening it shows the
> > > default value for half-life time is 5 seconds (not 15..)
> > >
> > > #sh interfaces dampening
> > >
> > > ATM0/0
> > > Flaps Penalty Supp ReuseTm HalfL ReuseV SuppV MaxSTm MaxP
> > > Restart
> > >
> > > 0 0 FALSE 0 5 1000 2000 20
> > > 16000 0
> > >
> > > #
> > >
> > > I did a test trying to dampen an interface after 3 flap in 30 seconds.
> >
> > > the config that I see work was with an half-life of 30 seconds and a
> > > suppress-threshold of 2500.
> > >
> > > I think if you want to dampen an interface after a certain number of
> > > flap in a determined time you have to set suppress thresold and also
> > > half-life time. wath do you think about ?
> > >
> > > thanks for any comments
> > >
> > > ______________________________________________________________________
> > > _
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> >
> > _______________________________________________________________________
> > Subscription information may be found at:
> > http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon Jan 09 2006 - 07:07:50 GMT-3