Nicolas Williams wrote:
> On Tue, May 29, 2007 at 01:53:48PM -0700, Garrett D'Amore wrote:
>
>> Kais Belgaied wrote:
>>
>>> Now, back the the symptoms (scope of this case): Each futile
>>> submission of a packet to be transmitted on a link down indicates a
>>> problem worth paying attention to. It could be uncovering a bug such
>>> as the above, or it could be transient race. I don't believe it is a
>>> bad practice from driver writers to adopt a defensive approach and
>>> log an error on every occurrence of the offense.
>>>
>> Yikes! That's a bad idea, if you mean doing it in syslog.
>>
>
> Why? Because of the risk of filling logs? If an interface is expected
> to flap a _lot_ then a knob to turn off logging about it would be
> useful. But I expect that such situations are so rare that any concern
> about filling logs is unrealistic.
>
Quite the contrary! If you run dhcp on multiple interfaces, then it
will periodically send probes out the network (or try at least). If you
have an interface marked up, but the cable is disconnected, then it will
flood the logs.
We do _not_ log an error in syslog every time a packet is dropped due to
collisions. Why should we do it for carrier_errors?
>
>> Of course, we _do_ log (collect) this information in the carrier_errors
>> kstats. Just like we do for all other network related errors.
>>
>
> Statistics != logging.
>
See above. Are carrier_errors somehow magical in that they get special
handling that no other errors should get?
-- Garrett
