Hi Jim.
Looks liekl calren is still being Interesting.
I've seen this before - on two different platforms. Neither of which were
the GSR that this seems to be from the name.
I've seen a 7206 with 12.1T, 12.1 and 12.2 amplify packets about 200
times. You could instantly fix the problem by disabling CEF. (The
problem does not occur in the 12.0S train.) Reenable CEF, and the
amplifcation resumes - but not straight away, and not for all IP
addresses transiting the router, and different IP addresses being
affected after different periods of time.
I've also seen 6500's MSFC II's with current code amplify packets about 4
times (as you see), in similar circumstances. No option to disable CEF
there.
However, both occurrences also had HSRP running, and removing that also
seemed to fix it.
(In first case, cisco saw it happen on my routers, but could not
reproduce in the lab, and basically gave up. I was OK with that as I
found it didn't occur in 12.0S, and I changed things so I didn't need the
features that were preventing me using that code in first place.
In second case -their still stumped and looking.)
Is HSRP involved at all here?
Cheers.
Steve Francis (ex-UCSB.)
Jim Warner wrote:
> When I ping to a particular router my reply packets are
> getting duplicated. It looks like:
>
> bash-2.03$ ping 137.164.12.33
> PING 137.164.12.33 (137.164.12.33): 56 data bytes
> 64 bytes from 137.164.12.33: icmp_seq=0 ttl=250 time=24.924 ms
> 64 bytes from 137.164.12.33: icmp_seq=0 ttl=250 time=25.013 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=0 ttl=248 time=25.247 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=0 ttl=248 time=25.698 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=1 ttl=250 time=24.084 ms
> 64 bytes from 137.164.12.33: icmp_seq=1 ttl=250 time=24.594 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=1 ttl=248 time=24.825 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=1 ttl=248 time=25.030 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=2 ttl=250 time=24.330 ms
> 64 bytes from 137.164.12.33: icmp_seq=2 ttl=250 time=24.544 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=2 ttl=248 time=24.750 ms (DUP!)
> 64 bytes from 137.164.12.33: icmp_seq=2 ttl=248 time=25.173 ms (DUP!)
>
> Not shown here, but this effect happens when I pass through this
> router. Anything beyond it is replicated.
>
> I have seen routing loops cause things that are similar, but
> not the same as this. But here the duplication is always the
> same. I think that my packets to the router are being replicated,
> and then each of the replies is getting replicated -- which is
> why I see four responses to everything I send.
>
> I am curious to hear what sorts of pathologies might result in
> this kind of packet copying. There is ATM in the path. Could
> it be the culprit? I am pretty sure that the equipment doing the
> packet copying is made by Cisco. The same way that I can't imagine
> that a bad packet would cause a fan to fail, I can't imagine how a
> packet could be copied a fixed number of times.
>
> I know this would be easier if there was a map of the topology.
> Is anyone willing to speculate without one?
>
> Thanks. -jim warner, UC santa cruz
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:56 EDT