[c-nsp] OSPF stability issues with 12.2SXI on 6500/Sup720?

[Justin M. Streiner]

On Tue, 3 Jan 2012, Chuck Church wrote:

> Any control plane policing?  We had a similar issue at last job, turned out
> a few of our OSPF sessions weren't placed in the above-best-effort class.
> Our IA people running a network scanner occasionally without telling us
> would cause many drops in the best effort class, causing OSPF session drops.
> After adding those few OSPF sessions to the right class, problem went away.
> This as with SXI3, and eventually SXI4.  Other than that, no issues.  That
> was with VSS mode, and lots of DFCs, if that makes a difference.

No control-plane policing.  I'm also in the process of comparing the 
config and hardware load-out on this particular 6509 to another one that's 
been running SXI7 for a month with no issues.  The only significant 
difference between the two is that the one that's running fine has a 
somewhat different complement of linecards and sees less traffic than the 
one we had to roll back, but the one we rolled back is not being 
saturated by any means.  SXI7 also did not seem to be hitting the 
CPU/memory/buffers on the 6509 we rolled back any harder than SXH5.

jms

> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Justin M. Streiner
> Sent: Tuesday, January 03, 2012 6:42 PM
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] OSPF stability issues with 12.2SXI on 6500/Sup720?
>
> Has anyone seen any serious stability issues with OSPFv2 in 12.2(33)SXI?
> I've tried both SXI6 and SXI7 Advanced Enterprise, and in both cases, after
> about two weeks of normal operation, OSPF sessions will start dropping on
> the affected switch, due to an expired dead timer.  Devices at the other
> ends of those adjacencies do not report them dropping for any odd reason,
> but they do report the adjacencies re-establishing.  I've seen no underlying
> interface-level errors.
>
> The issue appears to be specific to OSPFv2.  I have not seen any stability
> problems in OSPFv3 adjacencies on the same device.  It almost seems like
> OSPFv2 in 12.2(33)SXI could be mangling hellos, or simply stop sending them.
> Unfortunately I didn't have the time to investigate that in depth, during
> the issue I dealt with today.
>
> I backed off to 12.2(33)SXH5 Advanced IP Services and the problem went away.
> This is the second time an upgrade to 12.2SXI was done and had to be backed
> off.  I haven't found anything interesting in the release notes yet for
> 12.2SXH - 12.2SXJ that leads to an answer.  A cursory look through the bug
> toolkit also didn't turn up any really good matches.
>
> The overall configuration isn't terribly exotic.  The 6500 is dual-stacked
> - IPv4 IGP is OSPFv2 and IPv6 IGP is OSPFv3.  There is one VRF being handled
> by a separate OSPFv2 process in the box.  There are LDP and MSDP peers with
> other core devices on the backbone.  No BFD, and currently no authentication
> on the OSPFv2 neighbors.
>
> I have a case open with TAC on this, but I wanted to see if these symptoms
> line up with behavior that other people might have seen.
>
> jms
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>
>