[nsp] GEIP+ overruns/missed packets

From: Hank Nussbacher (hank@att.net.il)
Date: Mon Aug 06 2001 - 05:02:09 EDT


I have a GEIP+ card that keeps getting overruns. When looking inside the
VIP (sho cont g0/0/0) I see mant missed packets:

FX1000 Statistics (PA0)
   CRC error 0 Symbol error 0
   Missed Packets 34758106 Single Collision 0
   Excessive Coll 0 Multiple Coll 0
   Late Coll 0 Collision 0
   Defer 0 Receive Length 0
   Sequence Error 0 XON RX 0
   XON TX 1 XOFF RX 0
   XOFF TX 4 FC RX Unsupport 0
   Packet RX (64) 32526089 Packet RX (127) 0
   Packet RX (255) 0 Packet RX (511) 0
   Packet RX (1023) 0 Packet RX (1522) 0
   Good Packet RX 32526092 Broadcast RX 1
   Multicast RX 0 Good Packet TX 0
   Good Octets RX.H 0 Good Octets RX.L 2081669888
   Good Octets TX.H 0 Good Octets TX.L 679494
   RX No Buff 0 RX Undersize 0
   RX Fragment 0 RX Oversize 0
   RX Octets High 0 RX Octets Low 11222016
   TX Octets High 0 TX Octets Low 679814
   TX Packet 6364 RX Packet 67284211
   TX Broadcast 4 TX Multicast 907
   Packet TX (64) 5448 Packet TX (127) 0
   Packet TX (255) 0 Packet TX (511) 907
   Packet TX (1023) 4 Packet TX (1522) 0

What does missed packets mean? On show int I see overruns at 54Kpps on a
GEIP+ (VIP4-80)!

VIP-Slot0#sho int
GigabitEthernet0/0 is up, line protocol is up
   Hardware is WISEMAN, address is 00d0.7939.a800 (bia 00d0.7939.a800)
   MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec, rely 255/255, load 1/255
   Encapsulation ARPA, loopback not set
   Full-duplex mode, link type is autonegotiation, media type is SX
   output flow-control is on, input flow-control is on
   ARP type: ARPA, ARP Timeout 04:00:00
   Last input never, output never, output hang never
   Last clearing of "show interface" counters never
   Queueing strategy: fifo
   Output queue 0/40, 0 drops; input queue 0/75, 0 drops
   5 minute input rate 19999000 bits/sec, 54343 packets/sec
   5 minute output rate 0 bits/sec, 0 packets/sec
      1062330 packets input, 48867180 bytes, 0 no buffer
      Received 2 broadcasts, 0 runts, 0 giants, 0 throttles
      0 input errors, 0 CRC, 0 frame, 68038990 overrun, 0 ignored
      0 watchdog, 0 multicast, 0 pause input
      0 packets output, 0 bytes, 0 underruns
      0 output errors, 0 collisions, 2 interface resets
      0 babbles, 0 late collision, 0 deferred
      3 lost carrier, 0 no carrier, 4 pause output
      0 output buffer failures, 0 output buffers swapped out

The config is very simple:

ip flow-cache feature-accelerate
ip cef distributed
!
interface GigabitEthernet0/0/0
  ip address 192.168.100.1 255.255.255.0
  no ip directed-broadcast
  ip route-cache flow
  ip route-cache distributed
  load-interval 30
  negotiation auto

I tried removing route-cache flow and distributed. I tried removing ip
flow-cache as well.

I am running with:
IOS (tm) RSP Software (RSP-PV-M), Version 12.0(14)S2, EARLY DEPLOYMENT
RELEASE SOFTWARE (fc1)
Copyright (c) 1986-2001 by cisco Systems, Inc.
Compiled Fri 12-Jan-01 12:27 by pwade
Image text-base: 0x60010950, data-base: 0x60DE2000

ROM: System Bootstrap, Version 11.1(8)CA1, EARLY DEPLOYMENT RELEASE
SOFTWARE (fc1)
BOOTFLASH: RSP Software (RSP-BOOT-M), Version 12.0(14)S2, EARLY DEPLOYMENT
RELEASE SOFTWARE (fc1)

c751e1 uptime is 15 hours, 33 minutes
System returned to ROM by reload at 08:32:55 UTC Wed Dec 6 2000
System image file is "slot0:rsp-pv-mz_120-14_S2.bin"

cisco RSP4 (R5000) processor with 262144K/2072K bytes of memory.
R5000 CPU at 200Mhz, Implementation 35, Rev 2.1, 512KB L2 Cache
----------------------------------

Last question: when specifying exactly 100Kpps, there was no overruns, but
when we configed "ip route-cache distributed" on the GE interface, we
started to lose .3% packets. I always thought "ip route-cache distributed"
helps and doesn't hurt. Our testing shows the opposite. Why?

-Hank



This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:12:48 EDT