We help IT Professionals succeed at work.

Cisco 6509 SUP720 IPv4 CEF Entries

HunTelWebProgrammer
HunTelWebProgrammer used Ask the Experts™
on
I have a set of 3 Cisco 6509 routers acting as my border routers, each with a BGP session to a different provider.  I filter the BGP table coming in from each provider to not send me routes anything smaller than a /24 due to my memory constraints on a SUP720PFC3A.  this works great.  However, last night, I accidentally took the filter off one of my peers, and it filled up the TCAM Utilization to 100%.  I quickly applied the filter, and reset my peering session and the usage then went back down to 40-50%.  HOWEVER.  I still get messages in syslog about the tcam overflow, and i am having some throughput issues globally, and packet loss on my loopback interfaces.  how do i clear the memory problem short of rebooting the chassis?  
(i have had this issue before, and rebooting the chassis fixed the issue, but i dont really want to do that)  attached is a file with some helpful info.  6509-1.txt
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®

Author

Commented:
i found the command "clear ip cef epoch full"  but i am not sure if that is what i am looking for.

Commented:
Hello, as I can see from output, you have problem with part of TCAM, where NetFlow entries are stored

Netflow Resources
          TCAM utilization:       Module       Created      Failed       %Used
                                                  6             124271           0         94%
          ICAM utilization:       Module       Created      Failed       %Used
                                                  6                  0       17310          0%

If you need to decrease this number, consider to set more aggressive timers on NetFlow entries, so they can expire faster. For example:

mls aging fast time 16 threshold 16
mls aging long 64
mls aging normal 32

If you need higher capacity, you can upgrade your SUP720 to -XL version (256 000 entries) or buy a DFC card for your linecards.

Nevertheless, problem with drops on loopbacks and so should have nothing in common with this problem, as TCAM table is partitioned and only NetFlow part is exhausted.

Author

Commented:
exactly.. so that is not my problem... Im not sure if i made it clear or not above, but this router runs fine for years at a time.. there is no memory "leak"  the reason why it did it was because I made some changes to BGP and forgot to filter incoming prefixes.   the memory error i am getting has absolutley nothing to do with netflow resources, its talking about my ipV4 routes, that after I fix my bgp filter, is now sitting at 46 some percent.   I am thinking this is just some bug where once the router goes into software routing mode, it never comes out of it until a reboot is done.  I rebooted all 3 border routers tonight, and wala... fixed forever again until i screw up my BGP sessions :)

Hmm, so if the error you are receiving now is the one at the end of file then it obviously says that some traffic is being software switched, hence the perfromance issue. Which on a 6500 is a death sentence since it is underpowered in that regard.  Once you reach that state the only way to recover is to reload unfortunately. Here is a good link;

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a00804916e0.shtml#cb

harbor235 ;}

Author

Commented:
my CPU and ram usage didn't really get all that high, however it did increase some.. didnt even get over 20% but it was definatley struggling as you said!!  I rebooted them last night, and all is well.  thanks for the info!!

Author

Commented:
thanks again!