[c-nsp] 6704-10GE huge input drops (flushes)

[Antonio Soares]

The outputs related with the fabric don't show any drops.

What bothers me is the impact: how could the drops affect the bgp/isis
adjacencies to the point of bringing them down ?


Regards,

Antonio Soares, CCIE #18473 (R&S/SP)
amsoares at netcabo.pt
http://www.ccie18473.net



-----Original Message-----
From: Mack McBride [mailto:mack.mcbride at viawest.com] 
Sent: terça-feira, 7 de Maio de 2013 01:56
To: Antonio Soares; cisco-nsp at puck.nether.net
Subject: RE: [c-nsp] 6704-10GE huge input drops (flushes)

For what every reason the box could not forward packets properly.

This could be caused by congestion on the output queue of a completely
unrelated interface, except that the packet at the head of queue was
destined for that interface. This usually happens on floods destined for a
1G port coming in on a 10G port.

This can also be caused by backplane congestion.  But that is a much larger
topic.


LR Mack McBride
Network Architect

-----Original Message-----
From: cisco-nsp [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of
Antonio Soares
Sent: Monday, May 06, 2013 4:59 PM
To: cisco-nsp at puck.nether.net
Subject: [c-nsp] 6704-10GE huge input drops (flushes)

Hello group,

I'm trying to find out what caused this huge increase in the input drops on
a 6704-10GE line card. 6509 chassis with SUP720 running 12.2.18SXF16.

Before:
++++++++++++++++++++++++++++++++
cat6k#sh int te1/1 | inc drops
  Input queue: 1/2000/10310609/10310609 (size/max/drops/flushes); Total
output drops: 0 cat6k#
++++++++++++++++++++++++++++++++

After:
++++++++++++++++++++++++++++++++
cat6k#sh int te1/1 | inc drops
  Input queue: 0/2000/15863293/15863293 (size/max/drops/flushes); Total
output drops: 0 cat6k#
++++++++++++++++++++++++++++++++

This is a L3 interface is connected directly to a 12K (sip-601+spa-10ge).

I was thinking about bursts or micro-bursts but when this happened, the
bgp/isis sessions that go over this link went down.

The 6704-10GE card has low buffers compared with newer models like the 6708
or 6716:

6704 - 16MB per port, 2MB Rx, 14 Mb Tx
6708 - 256MB per port, 109MB Rx, 92MB Tx

But it makes some sense because the 6704 does full-rate and the 6708/6716
are oversubscribed (more buffering capacity needed).

The drops are flushes so it should mean something related with buffers. Or
maybe the old release that is running has some issues.


Any hints ?


Thanks.

Regards,

Antonio Soares, CCIE #18473 (R&S/SP)
amsoares at netcabo.pt
http://www.ccie18473.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/