2620 IP CEF lockups

[pgordemer]

I have a Cisco 2620 with 64MB memory (Max memory for the router model) running 12.3(19)ipbase. This router has been rock solid for a very long time. 3 months ago, we added a second T1 and using IP CEF to join them together (MLPPP is not supported by the ISP).

Regularly now, (Once a week) the router gets a semi lockup condition in that you can talk to the router, checks its configs, but BOTH serial ports are marked down. Shut and No Shut doesn't bring them back, you have to reboot the router.

I am wondering maybe with all the routes and CEF that the router is running out of memory. (Maybe i need something like no ip-route cache). I sniffed the network packets and there isn't anything unusual out there (no slammer, DOS, etc).

I have already swapped T1 cards on port interfaces out, have had the lines checked, etc..

Here is the config, it is very very basic. Hopefully someone will see something I am missing.


version 12.3
service timestamps debug datetime msec
service timestamps log datetime msec
service password-encryption
!
hostname arh_hq_internet
!
boot-start-marker
boot-end-marker
!
!
no aaa new-model
ip subnet-zero
ip cef
!
!
!
ip name-server 146.145.64.2
!
username xxx privilege 15 password 7 xxxxxxxxxxxxx
username xxxxxx privilege 15 password 7 xxxxxxxxxxx
!
!
class-map match-any VOIP
match ip dscp ef
match ip dscp af41
!
!
policy-map PRIORITY
class VOIP
priority 768
class class-default
fair-queue
random-detect dscp-based
!
!
!
interface FastEthernet0/0
description Ethernet to Sonicwall 2040
bandwidth 100000
ip address xxx.xxx.xxx.161 255.255.255.248
speed 100
full-duplex
!
interface Serial0/0
description T1 to ATX Circuit ID:HCGS307601NJ Jack J3
bandwidth 1536
ip address xxx.xxx.xxx.170 255.255.255.252
service-policy output PRIORITY
!
interface Serial0/1
description T1 to ATX Circuit ID:HGCS314067NJ Jack J1
bandwidth 1536
ip address xx.x.xxx.38 255.255.255.252
service-policy output PRIORITY
!
ip classless
ip route 0.0.0.0 0.0.0.0 Serial0/1
ip route 0.0.0.0 0.0.0.0 Serial0/0
no ip http server
!
dialer-list 1 protocol ip permit
!
line con 0
login local
line aux 0
line vty 0 4
exec-timeout 30 0
login local
!
!
end

Thanks
Phil Gordemer
ARH Associates

 

[vipergg]

I doubt you are running out of memory , you only need 32 minimum to run ipbase and you have twice that unless the memory is flakey , I would think it would take the whole box down if that was the case. Any line errors on the serials, or anything in the log ???

 

[pgordemer]

In all cases there are no line errors, and looking at show service modules, there are no alarms. In show int, the packets in and out are relatively evenly balanced considering I can do it per byte (VOIP gets cranky when you do it.

 

[burtsbees]

That is strange---sounds to be like some hardware...
The things I would do is take the cover off and clean all the dust and make sure the system is adequately cooled. I would actually run it for a while with the cover off.
Seems like a simple enough config...the other thing I would do is shut down one serial for 2 weeks, and only use one. Then if after 2 weeks it does NOT lock up, shut the other serial down, still only using one. If there is still no lockup issues, then go back to using both. Hopefully, it may just "go away"...
If it comes back during any of these scenarios, then at least you will have narrowed it down and perhaps know where to look next. I just bought a couple of 2620's with 64MB memory and 16 MB flash off of Ebay, one for $160 and another for 145. The memory and flash upgrades were $40 total, so make that about $200, but that's without serial interface modules. Good luck, my friend.

Burt

Tim

 

[3wsparky]

interesting that you mention ip cef in your problem

if you werent running ip cef and experiencing lockups then i could understand that as there is alot of overhead on the box when no using cef , but to be using cef and having this issue is confusing , would suspect hardware , maybe shut an interface for a period of time to see if it clears the fault , if so also swap the interfaces to check its not a backbone problem in the router.