[j-nsp] RE: juniper-nsp Digest, Vol 18, Issue 8

HULIN Emmanuel DRLD/ITP emmanuel.hulin at francetelecom.com
Fri May 14 04:12:17 EDT 2004


if your router increased the fabric drops, it is perharps if you received some unknown protocols for example ( cisco:cdp,udld,...).

Regards.

-----Message d'origine-----
De : juniper-nsp-bounces at puck.nether.net
[mailto:juniper-nsp-bounces at puck.nether.net]De la part de
juniper-nsp-request at puck.nether.net
Envoye : jeudi 13 mai 2004 18:01
A : juniper-nsp at puck.nether.net
Objet : juniper-nsp Digest, Vol 18, Issue 8


Send juniper-nsp mailing list submissions to
	juniper-nsp at puck.nether.net

To subscribe or unsubscribe via the World Wide Web, visit
	http://puck.nether.net/mailman/listinfo/juniper-nsp
or, via email, send a message with subject or body 'help' to
	juniper-nsp-request at puck.nether.net

You can reach the person managing the list at
	juniper-nsp-owner at puck.nether.net

When replying, please edit your Subject line so it is more specific
than "Re: Contents of juniper-nsp digest..."


Today's Topics:

   1. fabric drops ? (Jerome Fleury)
   2. RE: fabric drops ? (harry)


----------------------------------------------------------------------

Message: 1
Date: Wed, 12 May 2004 18:59:10 +0200
From: Jerome Fleury <jeje at jeje.org>
Subject: [j-nsp] fabric drops ?
To: juniper-nsp at puck.nether.net
Message-ID: <118843938.1084388350@[10.33.50.83]>
Content-Type: text/plain; charset=us-ascii

Hi,

does anyone know what fabric drops are ? (see last line)

we have this counter increasing on all of our routers, but one of them has much more than the
others.

Can someone point me on the exact definition of this ? Thanks.

jfl at RE0.BB1.COU> show pfe statistics traffic
PFE Traffic statistics:
        7887025183350 packets input  (1505803 packets/sec)
        7861203968641 packets output (1492453 packets/sec)

PFE Local Traffic statistics:
  675057223 local packets input
  675182639 local packets output
          0 software input high drops
        187 software input medium drops
          0 software input low drops
          0 software output drops
  393287161 hardware input drops

PFE Local Protocol statistics:
          0 hdlc keepalives
          0 atm oam
          0 fr lmi
   11617138 ppp lcp/ncp
    5507027 ospf hello
    3656359 rsvp hello
   19263171 isis iih

PFE Hardware Discard statistics:
          0 timeout
          0 truncated key
          0 bits to test
          0 data error
          0 stack underflow
          0 stack overflow
 4180034742 normal discard
          0 extended discard
      22104 invalid iif
          0 info cell drops
   61875452 fabric drops


--
Jerome Fleury


------------------------------

Message: 2
Date: Wed, 12 May 2004 10:46:49 -0700
From: "harry" <harry at juniper.net>
Subject: RE: [j-nsp] fabric drops ?
To: "'Jerome Fleury'" <jeje at jeje.org>, <juniper-nsp at puck.nether.net>
Message-ID: <006701c43849$1a311990$6401000a at jnpr.net>
Content-Type: text/plain;	charset="us-ascii"

I believe that these represent packets that were received by an ingress
T-series FPC and destined to egress on a different FPC. This traffic must
transit the T-series switch fabric. In this case it seems that the traffic
was dropped because the destination FPC was not able to service the traffic
in the confines of the T-series delay-bandwidth buffer (some 100
milliseconds, if not mistaken). Packets that are not serviced in this
delay-bandwidth time-period are simply overwritten by newer traffic.

I would suspect that you have an issue of more ingress traffic being
directed to egress on a given FPC/PIC/port than that FPC/PIC/port can
handle. This many to one situation can lead to fabric drops; why send
traffic over the fabric to an egress FPC, only to have it dropped by the
egress FPC because it is unable to handle the offered load? This may also be
an issue of incorrect CoS related settings, or perhaps the use of
fabric-override, which is rarely used.  

Some information can be found at:

http://www.juniper.net/techpubs/software/junos/junos63/swcmdref63/html/cos-m
onitor6.html#1137498

HTHs



> -----Original Message-----
> From: juniper-nsp-bounces at puck.nether.net 
> [mailto:juniper-nsp-bounces at puck.nether.net] On Behalf Of 
> Jerome Fleury
> Sent: Wednesday, May 12, 2004 9:59 AM
> To: juniper-nsp at puck.nether.net
> Subject: [j-nsp] fabric drops ?
> 
> 
> Hi,
> 
> does anyone know what fabric drops are ? (see last line)
> 
> we have this counter increasing on all of our routers, but 
> one of them has much more than the others.
> 
> Can someone point me on the exact definition of this ? Thanks.
> 
> jfl at RE0.BB1.COU> show pfe statistics traffic
> PFE Traffic statistics:
>         7887025183350 packets input  (1505803 packets/sec)
>         7861203968641 packets output (1492453 packets/sec)
> 
> PFE Local Traffic statistics:
>   675057223 local packets input
>   675182639 local packets output
>           0 software input high drops
>         187 software input medium drops
>           0 software input low drops
>           0 software output drops
>   393287161 hardware input drops
> 
> PFE Local Protocol statistics:
>           0 hdlc keepalives
>           0 atm oam
>           0 fr lmi
>    11617138 ppp lcp/ncp
>     5507027 ospf hello
>     3656359 rsvp hello
>    19263171 isis iih
> 
> PFE Hardware Discard statistics:
>           0 timeout
>           0 truncated key
>           0 bits to test
>           0 data error
>           0 stack underflow
>           0 stack overflow
>  4180034742 normal discard
>           0 extended discard
>       22104 invalid iif
>           0 info cell drops
>    61875452 fabric drops
> 
> 
> --
> Jerome Fleury
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net 
> http://puck.nether.net/mailman/listinfo/junipe> r-nsp
> 




------------------------------

_______________________________________________
juniper-nsp mailing list
juniper-nsp at puck.nether.net
http://puck.nether.net/mailman/listinfo/juniper-nsp


End of juniper-nsp Digest, Vol 18, Issue 8
******************************************




More information about the juniper-nsp mailing list