[c-nsp] necessity of nowadays

[quinn snyder]

> On 23 Mar 2016, at 06:22, Phil Mayers <p.mayers at imperial.ac.uk> wrote:
> 
> Agreed. UDLD was a net problem for us - I can't think of a single time it performed it's function, but can recall a handful of outages from false-positives.

in my case, as an (earlier) adopter of nexus 7000 and running 4.x, 5.0, and 5.1 code — i saw numerous cases where a supervisor switchover (during issu, for example) would cause the scheduler to eat itself, and in turn, cause control-plane protocols to drop.  most often, this was not a huge issue (more than some logs and small “blips”), but in the case of udld aggressive, i lost connectivity to the rest of the network, because udld was only showing down on the n7k side (not the far end).

lots of testing later — udld is only enabled in normal mode — if at all.  i agree that udld is more trouble than its worth unless specific corner cases are encountered.

q.

--
quinn snyder | snyderq at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 842 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://puck.nether.net/pipermail/cisco-nsp/attachments/20160323/29ce90c4/attachment.sig>