[c-nsp] Cisco 6500 experiencing %CPU_MONITOR-SP-6-NOT_HEARD

[Phil Mayers]

On 17/06/10 10:00, Youssef Bengelloun-Zahr wrote:
> Hello again Jeroen,
>
> Looked at the Cisco page regarding resolved caveats regarding
> 12.2(33)SXH rebuilds, only found entry CSCsm84267 about
> %CPU_MONITOR-SP-6-NOT_HEARD but the symptoms doesn't match our case.
>
> Still digging.

Well, my comment was related to whether SXI worked for the feature set 
you asked about, rather than your original %CPU_MONITOR issue.

Having said that: I'm fairly sure the SXH and SXI have significant 
differences in the way IPC happens, some of which is related to ISSU, 
some of which seems to be more aggressive monitoring (or even just 
logging of such)

We've had %XDR_* errors where standby supervisors "go away"; TAC claimed 
it was a hardware fault, but to my mind that's not credible, especially 
since the hardware passes full GOLD diags in our test chassis, and 
various combinations of reboots and upgrades made it "go away".

[Having said that: TAC basically say "duh RMA the sup" about 50% of the 
time in cases like this. Some engineers seem to think memory corruption 
twice in a row means hardware fault, because of course memory corruption 
can't come from a software bug. We end up having to explain "memory 
santiy debug" to them, and the fact that corruption might not have been 
caused by the process that triggered the traceback...]

In short; I think all this infrastructure changed from SXF to SXH/SXI 
and I think the early versions had bugs. I know from past experience 
that Cisco don't necessarily expose all bugs fixed in the bug toolkit, 
so there may be a bug that matches yours and I'd just upgrade and see if 
it stops.