[c-nsp] 7206 crashing when enabling LLQ on Frame-relay interface(s)

[Rolf Mendelsohn]

Hi Everyone,

Thanks to those who replied off-list. To summarise 12.2SB is a new train 
focusing on ISG and I should not be running it anyway.

For those who are interested I also found the bug under bug toolkit:

CSCea72991 Bug Details
   
  Headline   E1/T1 on VIP: %SYS-2-QCOUNT: Bad deqeueue
  Product   IOS
  Feature   QOS   Duplicate of   
  Severity   2  Severity help  Status   Resolved  Status help
  First Found-in Version   12.2(16)   All affected versions   First Fixed-in 
Version   12.0(23)S04, 12.2(17.7)S, 12.0(25.3)S01, 12.3(2.1), 12.2(17.8), 
12.1(19.2)E, 12.1(19)E02, 12.3(3.1)T, 12.0(24)S03, 12.2(15)T07, 12.3(2.3)B, 
12.2(15)ZK01, 12.0(25)S02, 12.0(25)SX02, 12.0(27)SV, 12.3(7)XI  Version help
  Release Notes
 
Symptoms
Following error may be reported on Cisco 7500 Series Router with an E1 or T1 
controller on a Versatile Interface Processor (VIP) card:

 %SYS-2-QCOUNT: Bad deqeueue 60B71288 count 0
 -Process= "<interrupt level>", ipl= 1
 -Traceback=

Conditions
Service-policy is applied to a channelized interface directly or indirectly 
(via
Multilink interface or Frame-Relay map-class). 

Under this condition, any of following actions triggers the defect:

 - reconfiguring a channel-group
 - a flap of the E1 or T1 controller
 - after reconfiguring the service-policy command on a
 - physical interface
 - Multilink interface which has a channelized member interface
 - Frame-relay map-class which is applied to a channelized interface

The defect is persistent under listed conditions.

Workaround
 - when reconfiguring a channel-group, shut down the interface beforehand

 - when reconfiguring a service-policy command applied directly to a physical 
interface, shut down the interface beforehand
 
 - when reconfiguring a service-policy command applied to a Multilink 
interface,
shut down the member interface(s) beforehand
 
 - when reconfiguring a service-policy command applied to a Frame-relay 
map-class, remove the map-class from the interface beforehand

From: 
http://www.cisco.com/cgi-bin/Support/Bugtool/onebug.pl?bugid=CSCea72991&cco_product=IOS&fset=&swver=&keyw=%25SYS-2-QCOUNT&target=&train=

cheers
/rolf


On Sunday 11 February 2007 20:21, Rolf Mendelsohn wrote:
> Hi Guys,
>
> I have a 7200 doing a bit of MPLS / routing, frame-relay FR.12
> fragmentation and now I wanted to add LLQ for prioritising traffic.
>
> However the router keeps crashing with traceback errors, i have tested on
> 12.2.28-SB4 and 12.2.28SB6
>
> I was wondering if anybody had come across the following:
>
> Feb 11 17:52:59.301 UTC: %SYS-2-QCOUNT: Bad deqeueue 63EF09D8 count 1
> -Process= "<interrupt level>", ipl= 1
> -Traceback= 60888024 60A0FF98 60A269C4 601B24BC 6051979C 6051CCE8 6051CFB4
> Feb 11 17:52:59.305 UTC: %SYS-2-QCOUNT: Bad deqeueue 63EF09D8 count 0
> -Process= "<interrupt level>", ipl= 1
> -Traceback= 60888024 60A0FF98 60A269C4 601B24BC 6051979C 6051CCE8 6051CFB4
> pe1.wdh1.na#
> Feb 11 17:53:00.641 UTC: %SYS-2-QCOUNT: Bad deqeueue 63EF09D8 count 0
> -Process= "<interrupt level>", ipl= 1
> -Traceback= 60888024 601206D4 6069814C 606CF504 605BE7C8 605BCFD4 6191D28C
> 601AF678 6051979C 6051CCE8 6051CFB4
>
> I will try and open a TAC case tommorrow, but am really in a hurry to try
> and get this finished.
>
> If the problem is frame-relay related then perhaps i should swap to
> Mulitilink PPP?
>
>
> The config is below:
>
> policy-map mpls-qos
>  description Set a General Backbone QoS policy
>   class qos1
>     priority percent 20
>   class qos2
>     bandwidth percent 30
>   class class-default
>     fair-queue
>
> class-map match-any qos1
>   description Critical Class
>   match mpls experimental topmost 5  6  7
>   match ip precedence 5  6  7
> class-map match-any qos2
>   description Priority class
>   match mpls experimental topmost 1  2  3  4
>   match ip precedence 1  2  3  4
>
> pe1.wdh1.na#sh mpls interfaces
> Interface              IP            Tunnel   BGP Static Operational
> FastEthernet0/1        Yes (ldp)     No       No  No     Yes
> Serial5/2              Yes (ldp)     No       No  No     Yes
> Serial6/0:0            Yes (ldp)     No       No  No     Yes
> Serial3/0:2.1          Yes (ldp)     No       No  No     Yes
> Serial3/0:5.1          Yes (ldp)     No       No  No     Yes
> Serial3/0:6.1          Yes (ldp)     No       No  No     Yes
> Serial3/0:7.1          Yes (ldp)     No       No  No     Yes
> Serial3/0:8.1          Yes (ldp)     No       No  No     Yes
> Serial3/0:10.1         Yes (ldp)     No       No  No     Yes
> Serial3/2:2.1          Yes (ldp)     No       No  No     Yes
> Serial3/5:0.1          Yes (ldp)     No       No  No     Yes
>
>
> I Have a couple of interfaces running MPLS, configured as such:
>
> interface Serial3/0:2
>  description 128k  Backbone link to Town B PoP
>  mtu 1516
>  bandwidth 128
>  no ip address
>  encapsulation frame-relay IETF
>  frame-relay lmi-type ansi
>  frame-relay intf-type dce
>  frame-relay fragment 192 end-to-end
> end
> !
> interface Serial3/0:2.1 point-to-point
>  description 128k  Backbone link to Town A PoP
>  bandwidth 128
>  ip address X.Y.Z.71 255.255.255.254
>  mpls ip
>  no cdp enable
>  frame-relay interface-dlci 206
> end
>
> cheers
> /rolf