Re: [c-nsp] SXI leaks (was: Netflow on SUP720-3BXL)
Hi, On Tue, Mar 17, 2009 at 11:32:01AM -0500, Brandon Ewing wrote: On Mon, Mar 16, 2009 at 04:55:04PM +0100, Gert Doering wrote: This is on a peering point + upstream provider router, with full IPv4 and IPv6 BGP (unicast only). SXI non-modular, advanced ip services. We lose about 2-4 Mbyte of free memory per day, which goes into holding for the BGP Router process. How are you monitoring this? Is there an OID that's easily watchable to keep an eye on this situation? I have an SXI box carrying full tables from multiple providers, so I'm definitely getting some churn on the box. There's an OID for free memory, out of CISCO-MEMORY-POOL.mib # query_base is the starting-point of CISCO-MEMORY-POOL.mib my $query_base = .1.3.6.1.4.1.9.9.48.1.1.1; (we just walk the whole tree and sort out which pool is what) Which process is holding the memory is grabbed by a script that runs once per day and runs show proc mem via rsh. Not pretty, but does the job. gert -- USENET is *not* the non-clickable part of WWW! //www.muc.de/~gert/ Gert Doering - Munich, Germany g...@greenie.muc.de fax: +49-89-35655025g...@net.informatik.tu-muenchen.de ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
[c-nsp] SXI leaks (was: Netflow on SUP720-3BXL)
(SXI has slow memory leaks in BGP, at least for us. Cisco case has been opened, but hasn't proceeded anywhere yet). Interesting. Do you have any details you can share? ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] SXI leaks (was: Netflow on SUP720-3BXL)
Hi, On Mon, Mar 16, 2009 at 03:15:01PM +, Phil Mayers wrote: (SXI has slow memory leaks in BGP, at least for us. Cisco case has been opened, but hasn't proceeded anywhere yet). Interesting. Do you have any details you can share? This is on a peering point + upstream provider router, with full IPv4 and IPv6 BGP (unicast only). SXI non-modular, advanced ip services. We lose about 2-4 Mbyte of free memory per day, which goes into holding for the BGP Router process. It seems to be related to churn - we have another router running SXI, and that one is used at the network edge with only about 500 BGP prefixes, and nearly no churn. That one has no (noticeable) memory leak. TAC Case# is SR 610821739. (... maybe we should really go for SXI modular... - just restart the BGP Router every 2 months, and reclaim all this memory without a 10-minute reboot... and maybe even get a bugfixed BGP-process without requiring a reboot). gert -- USENET is *not* the non-clickable part of WWW! //www.muc.de/~gert/ Gert Doering - Munich, Germany g...@greenie.muc.de fax: +49-89-35655025g...@net.informatik.tu-muenchen.de pgpxRv5ev5zHH.pgp Description: PGP signature ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
Re: [c-nsp] SXI leaks (was: Netflow on SUP720-3BXL)
Hi, On Mon, Mar 16, 2009 at 11:09:48AM -0500, Murphy, William wrote: Thanks for the heads up... Are any of the maintenance releases of SXH also affected by this memory leak? We've not seen this in SXH3a (and you don't want to use SXH3 due to BGP ghost bugs). I have not tested SXH4. gert -- USENET is *not* the non-clickable part of WWW! //www.muc.de/~gert/ Gert Doering - Munich, Germany g...@greenie.muc.de fax: +49-89-35655025g...@net.informatik.tu-muenchen.de pgpwNKx89e9dd.pgp Description: PGP signature ___ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/