[c-nsp] ME6524 dying

[Bernhard Schmidt]

Bernhard Schmidt <berni at birkenwald.de> wrote:

> we have a ME6524-GT series in a remote location that keeps dying on us
> every couple of weeks. Symptoms are:
>
> * All routing protocols and LLDP time out on our side
> * Physically the box looks fine, Status LED and even the link LEDs are
>   still green
>   - I cannot really say something about the physical link status, since
>     the link in question does not forward link-state to our side
> * Console is dead, no reaction at all
>   - we do not have any OOB equipment on location yet, so I cannot tell
>     whether it emitted a dying gasp
> * CPU, Memory, all look normal up to five minutes before the crash
> * remote syslog is enabled, nothing visible in it
> * no crashdump on any file system
> * looks fine after power cycle, no warnings anywhere
>
> The box is not doing much at all, OSPFv2/v3, MPLS LDP and BGP towards
> our core and one partial transit, maybe 4000 prefixes, 200 Mbps of
> traffic.
>
> Up to now we have been running 12.2(33)SXI5 Adv.IP non-modular, I have
> now downgraded to SXI4. 

For the record, downgrading to SXI4 did not fix it. We have received an
off-list reply from someone telling me that his ME6524s exhibited the
same symptoms when running NDE. Disabled it, box has been stable so far. 

Bernhard