We have a 6509 running in native IOS mode that will spontaneously lose all of its OSPF and BGP routes. A 'show ip route' reveals all of the directly connected subnets (which are not reachable) and no other routes. The only solution to the problem is to reload the entire box. Any suggestions? This 6509 is our primary gateway router, so needless to say when it goes down, people scream. Cisco has been thoroughly unhelpful to this point - the TAC officer's only suggestion was to turn off OSPF and BGP. Yeah, right.
I do not see how turning off OSPF and BGP is even work-around much less a suggestion.
Couple of questions.
1. Software version
2. Due the neighbors drop prior to the routes being removed (i.e. do you see the BGP neighbors drop?)
3. Did you check for physical layer issues to the neighbor?
4. Is STP stable?
5. Link up/down occuring?
6. Any tracebacks or other console messages.
7. What type of line cards?
Could be hardware, but I imply software over hardware.
2. No. They just disappear and become unreachable.
3. Yes. No problems.
7. One SUP-II blade, one 24-port 100Mb-FX blade, one 16-port 1000Mb GIBC blade, 2-port Flex-WAN Adapter with serial and ATM modules, and one SFM blade.
During your outage situation have you collected any other type debug or HW specific information "show module all, etc". Possibly you have a module (FlexWan or other) that is going into a down state when this happens. If you can provide some detailed "show tech" information then I am sure that someone on this or other alias will be able to help you much better. With the limited information you've provided it would be pretty unlikely anyone could do much more than guess. Please provide better information to the nature of your problem.
I quick suggestion would to noted the behavior without the DFC card in the chassis. Are OSPF and BGP neighbors over the FlexWan?
first things first somebody said take the sh tech-support after reloading u cannot take a tech-support as that will be useless coz after reload everything comes back to normal u have already mentioned that so nothing will be available and when it is in hang state u will not be able to issue session 15
coz that a very low priority process pls give shecduler allocate to allocate cpu time for your seesion 15 during the hang state configure a crashdump are all the cards seated properly confirm that
and yes the tac guy is right u r running ospf and bgp on a 6509 which a switch not a router if my knowledge goes right its not made to handle ospf bgp i have never seen such a configurationg first of all it would be a privelege pls show me your configuration :)
You certainly got Cisco's attention i see.::))
It sounds or in fact is the box loosing OSPF neighbours and BGP peers, as suggested by previous posting a sh tech would help and maybe more a sh log.
Have you tried to shut down interfaces to get peers etc back up, or was a reload the only suggestion, if it was it almost sounds like a memory issue of some nature or definately a IOS bug if H/W has been proven to be stable.
How many routes do you hold with respect to BGP and OSPF and have you noticed a pattern or is it completely random.??
I see there are some issues (Bug CSCdx04363) with IP and Gigabit connectivity which is resolved in the E1 release.. i have not a done a bug search but just a quck peruse and there are a few BGP/OSPF bugs but your problem does not seem to be specifically related to OSPF and BGP as all routes dissapear so it seems more like a mem/IOS issue.
I have experienced the same problem. After disabling WCCP that redirects http traffic to a cache the problem dissapeared. I had to replace the chassi and supervisor card to be able to enable WCCP again.
I hope this will help you !
is your problem fixed ?
I tried to find the case you opened with the TAC but could not find anything related to this issue.