Prev: PIX PDM problem
Next: PPPoE Server - WIth Nat
From: Rob on 10 Nov 2009 04:38 mikeyb <bamikeb(a)hotmail.com> wrote: > >> >> In famous words of Cisco TAC - "Upgrade to latest mainline and call us >> back!" :-) >> >> Andrey. > > I suspected this might be needed, I assume this would be more economic > that getting a new router, but what is the downside to putting new > software on old kit? Usually a new IOS version won't fit in an old router without a memory upgrade. I have had very strange problems with CEF as well, when combined with dialers (unfortunately required for ADSL with PPPoA) and also with policy routing.
From: mikeyb on 10 Nov 2009 10:43 > You might want to check your CEF adjacencies after the stream stops. > > Dan Lanciani > ddl(a)danlan.*com Thanks Dan, I've checked the adjacencies and they ok (to me) . IPs connected to the right interfaces. The only thing I found using debug ip cef drops was lots of drops on the loopback interface (in the config to stop vpn traffic being static NAT'd). I don't think this is my problem though. Mike
From: bod43 on 10 Nov 2009 13:17 On 10 Nov, 09:38, Rob <nom...(a)example.com> wrote: > mikeyb <bami...(a)hotmail.com> wrote: > > >> In famous words of Cisco TAC - "Upgrade to latest mainline and call us > >> back!" :-) > > >> Andrey. > > > I suspected this might be needed, I assume this would be more economic > > that getting a new router, but what is the downside to putting new > > software on old kit? > > Usually a new IOS version won't fit in an old router without a memory > upgrade. Miraculously I seemed to have guessed the correct feature set and so you can see above the memory requirements. Image Name c1700-k9o3sy7-mz.124-25b.bin DRAM / Min Flash 96 / 32 Same as for 12.3T. You have enough RAM and Flash. Of course 12.4 mainline is basically the last development of 12.3T but now with 25 and more rounds of bug fixes:-) or :-(. I can recall doing PBR to a loopback to avoid NAT but we stopped years ago and did it differently. I did not do much static NAT and can't recall the details now. Not seen that for years anyway. http://www.cisco.com/en/US/docs/ios/12_2t/12_2t4/feature/guide/ftnatrt.html NAT - Ability to Use Route Maps with Static Translations 12.2(4)T This feature was introduced. So it looks slike you could remove the PBR if you preferred. It always seemed like a horrible kludge to me anyway.
From: Rob on 10 Nov 2009 13:26 bod43 <Bod43(a)hotmail.co.uk> wrote: > I can recall doing PBR to a loopback > to avoid NAT but we stopped years ago and > did it differently. I did not do much static NAT > and can't recall the details now. Not seen that for > years anyway. > > http://www.cisco.com/en/US/docs/ios/12_2t/12_2t4/feature/guide/ftnatrt.html > NAT - Ability to Use Route Maps with Static Translations > 12.2(4)T This feature was introduced. > > So it looks slike you could remove the PBR if you > preferred. It always seemed like a horrible kludge to > me anyway. We had to use PBR because we had two ADSL interfaces to internet, each with source address filtering. As you cannot randomly send traffic out to the ADSL in this case, as happens when you set two default routes, we used PBR with a loopback interface for all the outbound traffic. (selecting the proper ADSL interface based on the source address of the traffic) This worked OK, but not with CEF. Now the ADSL lines are retired and replaced by a single fiber, the problem is gone and CEF is now enabled on the router. IOS is 12.4(5a), has been updated several times but it never fixed the issue.
From: Dan Lanciani on 11 Nov 2009 00:06
In article <77a88baf-3614-4319-b7e0-ff53037e58f5(a)p35g2000yqh.googlegroups.com>, bamikeb(a)hotmail.com (mikeyb) writes: | > You might want to check your CEF adjacencies after the stream stops. | > | > Dan Lanciani | > ddl(a)danlan.*com | Thanks Dan, I've checked the adjacencies and they ok (to me) . IPs | connected to the right interfaces. The only thing I found using debug | ip cef drops was lots of drops on the loopback interface (in the | config to stop vpn traffic being static NAT'd). I don't think this is | my problem though. Well, drops are bad if you need the packets routed. :) Do the drops start as soon as CEF is enabled or after the problem occurs? Did you try disabling CEF on the loopback interface (only)? Dan Lanciani ddl(a)danlan.*com |