Re: bgpctl sh rib mem
On Sun, Sep 22, 2013 at 01:37:46PM +0200, Denis Fondras wrote: Hello, I have an OpenBGPd router with OpenBSD5.3, peering with two remote routers (one for v4 and one for v6). I expect my peers to send me only valid routes. Is there a way to show every prefixes sent by a peer, even invalid ones ? I know of bgpctl show rib but I'd like to understand what this line means : 27808 prefix entries using 869K of memory For me it looks like half of the prefixes I receive are invalid (in memory bu not in RIB) and I'd like to understand why. Thank you in advance, Denis $ bgpctl sh rib mem RDE memory statistics 15 IPv4 unicast network entries using 360B of memory 13889 IPv6 unicast network entries using 488K of memory 27807 rib entries using 869K of memory 27808 prefix entries using 869K of memory 8364 BGP path attribute entries using 621K of memory 8018 BGP AS-PATH attribute entries using 265K of memory, and holding 8364 references 1013 BGP attributes entries using 23.7K of memory and holding 1534 references 1012 BGP attributes using 7.9K of memory RIB using 3.1M of memory $ bgpctl show rib inet6 | wc -l 13902 $ bgpctl sho rib inet | wc -l 19 You most probably run with softreconf in and therefore there is to prefix entries for each path. bgpctl show tables - to show the tables which can be used bgpctl show rib in - to show the Adj-Rib-In bgpctl show rib table XXX - show the table XXX Invalid pathes are shown in either bgpctl show rib or in bgpctl show rib in -- :wq Claudio
Re: bgpctl sh rib mem
Hi Claudio, You most probably run with softreconf in and therefore there is to prefix entries for each path. This is it, disabling soft-reconfiguration lowers memory usage. Thank you very much for this accurate answer :) Denis
bgpctl sh rib mem
Hello, I have an OpenBGPd router with OpenBSD5.3, peering with two remote routers (one for v4 and one for v6). I expect my peers to send me only valid routes. Is there a way to show every prefixes sent by a peer, even invalid ones ? I know of bgpctl show rib but I'd like to understand what this line means : 27808 prefix entries using 869K of memory For me it looks like half of the prefixes I receive are invalid (in memory bu not in RIB) and I'd like to understand why. Thank you in advance, Denis $ bgpctl sh rib mem RDE memory statistics 15 IPv4 unicast network entries using 360B of memory 13889 IPv6 unicast network entries using 488K of memory 27807 rib entries using 869K of memory 27808 prefix entries using 869K of memory 8364 BGP path attribute entries using 621K of memory 8018 BGP AS-PATH attribute entries using 265K of memory, and holding 8364 references 1013 BGP attributes entries using 23.7K of memory and holding 1534 references 1012 BGP attributes using 7.9K of memory RIB using 3.1M of memory $ bgpctl show rib inet6 | wc -l 13902 $ bgpctl sho rib inet | wc -l 19
Re: bgpctl sh rib mem
On 2013-09-22, Denis Fondras open...@ledeuns.net wrote: Hello, I have an OpenBGPd router with OpenBSD5.3, peering with two remote routers (one for v4 and one for v6). I expect my peers to send me only valid routes. Is there a way to show every prefixes sent by a peer, even invalid ones ? IIRC there isn't.. I know of bgpctl show rib but I'd like to understand what this line means : 27808 prefix entries using 869K of memory For me it looks like half of the prefixes I receive are invalid (in memory bu not in RIB) and I'd like to understand why. The most likely reasons are invalid nexthops (bgpctl sh nex) or that the paths are dropped by your filter rules.
Re: bgpctl sh rib mem
Thank you very much Stuart. The most likely reasons are invalid nexthops (bgpctl sh nex) or that the paths are dropped by your filter rules. I have no explicit filter rules in bgpd.conf and bgpctl sh nex shows only valid nexthops. I restarted the daemon but no changes in memory usage. Denis
Re: bgpctl sh rib mem
On 2013-09-22, Denis Fondras open...@ledeuns.net wrote: Thank you very much Stuart. The most likely reasons are invalid nexthops (bgpctl sh nex) or that the paths are dropped by your filter rules. I have no explicit filter rules in bgpd.conf and bgpctl sh nex shows only valid nexthops. None at all? You should at least be blocking the private-use networks listed in RFC5735, and prefixes longer than /24 etc. I restarted the daemon but no changes in memory usage. It might be easier to help if we could see (sanitized, if necessary) bgpd.conf.