[c-nsp] watchdog timeout - nmi reset

[Mark Tinka]

Hi all.

We've had a bit of bad luck lately with a couple of NPE-G1's 
suddenly reloading due watchdog timeouts.

In all cases, we've been running 12.2SRC (first SRC1, and 
currently SRC2). Without any crashinfo generated from the 
reload, Cisco say this points to a hardware problem.

We initially experienced this on an NPE-G1 built in 2003 
(the chassis might have been built about the same time 
also). But then it also affected NPE-G1's built in 2005, as 
well as 2007.

We swapped out one of them that has been rebooting more 
frequently (once every 2 months) with a 2007-model NPE-G1. 
This just failed a few days back, same reason.

This morning, yet another 2007-model NPE-G1 also experienced 
the same problem. This one had never done this before. It 
also is installed in a 2007-model chassis.

The following is consistent:

* The watchdog timeout reset is affecting only our NPE-G1's.
* All NPE-G2's and 7201's, running SRC2, are not affected.
* It affects both old and new NPE-G1's.
* It affects both old and new chassis'.
* All routers are running 12.2(33)SRC2.

We're going to open another case with TAC on this, but I 
feel this is going to be drawn out.

It would have been easier if this affected either ONLY the 
old model NPE-G1's, or the new model NPE-G1's; because then 
we could either chalk it down to old boards or a bad batch 
(the 2007 models were all built to the same order).

But since this is affecting both old and new, and the 
information suggests it's not software-related, it gets 
tricky.

Aside from software, the only other thing that unites both 
the old and new chassis'/processors is PA-2FE-TX cards we 
bought for both the old and new models.

Suffice it to say that before the older NPE-G1's were 
running SRC (they either run 12.3 mainline or 12.2S), we 
didn't see this issue.

Appreciate any thoughts here.

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/20081106/5403470f/attachment.bin>