[c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Peter Rathlev
Good morning all, I'm stumped researching a slightly overloaded Supervisor 720 on one of our aggregation devices. I've discovered that an access-list applied to a SVI means denied packets are punted to the CPU. There's no log statement. The packets have no IP options, TTL=64, DSCP=0x28 and frame

Re: [c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Peter Rathlev
On Wed, 2012-08-29 at 11:17 +0200, Peter Rathlev wrote: I'm stumped researching a slightly overloaded Supervisor 720 on one of our aggregation devices. ... Forgot to mention platform: WS-SUP720-3B revision 5.2 running 12.2(33)SXI1 Advanced IP Services. Traffic arrives tagged on a WS-X6724-SFP.

Re: [c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Brian Turnbow
[mailto:cisco-nsp- boun...@puck.nether.net] On Behalf Of Peter Rathlev Sent: mercoledì 29 agosto 2012 11:18 To: cisco-nsp Subject: [c-nsp] Sup720 SVI ACL deny punted? (no logging) Good morning all, I'm stumped researching a slightly overloaded Supervisor 720 on one of our aggregation devices

Re: [c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Jared Mauch
What are your mls rate limiters set for, including the no-route one? Jared Mauch On Aug 29, 2012, at 5:17 AM, Peter Rathlev pe...@rathlev.dk wrote: Good morning all, I'm stumped researching a slightly overloaded Supervisor 720 on one of our aggregation devices. I've discovered that an

Re: [c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Peter Rathlev
On Wed, 2012-08-29 at 11:09 +, Brian Turnbow wrote: 1 to generate an ip unreachable ? try disabling them on the SVI Ahh, interesting idea. We have an ACL drop rate-limiter in place: mls rate-limit unicast ip icmp unreachable acl-drop 200 10 When replacing this with ... acl-drop 0 the

Re: [c-nsp] Sup720 SVI ACL deny punted? (no logging)

2012-08-29 Thread Peter Rathlev
On Wed, 2012-08-29 at 15:22 +0100, Phil Mayers wrote: On 29/08/12 14:35, Peter Rathlev wrote: If we know that the punting is limited to 200 pps it shouldn't matter too much. I've tried simply removing the ACL to see if the CPU overload disappears. But why would 200 pps even start making it