[c-nsp] IS-IS LSP Generation/Expiry + Database Optimization - Issue - Update!

[Mark Tinka]

Hi all.

Just an update for the archives and folk interested:

TAC came back with an explanation (and solution) to this 
issue.

The issue is that pseudonode link state PDU's don't have MT-
ID's for v6, while non-pseudonodes do.

The iSPF bug in the code doesn't handle pseudonode LSP 
changes for non-base topologies correctly.

Typically, during an iSPF run, changes to the pseudonode are 
applied to all topologies even though the pseudonode LSP's, 
themselves, don't contain any MT-ID's for v6. However, this 
bug creates a situation where, during an iSPF calculation 
for the v6 topology (MT-IPv6), iSPF would only calculate for 
changes that contain MT-ID's. As such, it would skip any 
changes for the pseudonode, leading to the incorrect SPF 
result.

The workarounds: disable multi-topologies and run a single 
topology or disable iSPF (both iSPF and multi-topologies 
should be enabled for the issue to present).

TAC say a fix for this issue will be available in the next 
release of SRD as well as SRC5.

In our case, our pseudonodes in our main PoP are 
6500/SUP720-3BXL's running SX*. However, as concerns these 
switches themselves, this isn't a problem as SX* doesn't 
support iSPF for v6. TAC did mention, though, that if/when 
SX* does support iSPF for v6, this fix will be incorporated.

Cheers,

Mark.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: This is a digitally signed message part.
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20090417/632ca62c/attachment-0001.bin>