sk0: watchdog timeout

[Vladimir Dvorak]

Danial Thom wrote:

>--- Vladimir Dvorak <dvorakv at vdsoft.org> wrote:
>
>  
>
>>Bernhard Fischer wrote:
>>
>>    
>>
>>>>>If you change hardware settings, you should
>>>>>          
>>>>>
>>also maintain the same
>>    
>>
>>>>>settings on both ends of the wire, i.e. at
>>>>>          
>>>>>
>>the computer *and* the
>>    
>>
>>>>>ethernet switch.
>>>>>     
>>>>>
>>>>>          
>>>>>
>>>>[SNIP]
>>>>
>>>>I just forced it to use 100baseTX /
>>>>full-duplex which I think was used before I
>>>>        
>>>>
>>forced it as well.
>>    
>>
>>>>   
>>>>
>>>>        
>>>>
>>>That's exactly what I ment. If you force your
>>>      
>>>
>>sk0 to 100-full do the same with 
>>    
>>
>>>your switch. If you set sk0 to auto-config --
>>>      
>>>
>>do it with your switch.
>>    
>>
>>>Watch your interface-counters: 
>>>
>>>netstat -I sk0 -b -d -t
>>>
>>>If the hardware (cards, cabeling, plugs, ...)
>>>      
>>>
>>is ok, there shouldn't be any 
>>    
>>
>>>errors for days or even weeks!
>>>
>>>Regards,
>>>bh
>>> 
>>>
>>>      
>>>
>>Hi Bernhard,
>>
>>thanks for your message. I suppose we all know
>>something about
>>networking, but this is hardware/driver related
>>problem ( with the
>>highest probability ). I used my server in two
>>different environments -
>>at home and now at serverhouse. No difference.
>>Still getting (from time to time ):
>>
>>sk0: watchdog timeout
>>sk0: link state changed to DOWN
>>sk0: watchdog timeout
>>sk0: link state changed to UP
>>sk0: watchdog timeout
>>sk0: link state changed to DOWN
>>sk0: watchdog timeout
>>sk0: link state changed to UP
>>sk0: watchdog timeout
>>sk0: link state changed to DOWN
>>
>>messages. But server still running. I am out of
>>ideas how can we solve
>>our problem and I am about to buy new network
>>card, because this
>>stressfull situation is not good. ;-)
>>
>>Vladimir
>>    
>>
>
>There are generally 2 cases in which you'll get a
>watchdog timeout message on a network card
>driver. 
>
>1) Your machine is in livelock (ie overrun)
>2) There is a problem with the interrupts
>
>Clearly you should know if 1) was occurring. With
>2), it could be a driver or hardware problem or
>both.
>
>DT
>
>  
>
Hi,

thanks for reply. The first case ( overrun ) is impossible, the server
has (for now) very low traffic ( load is 0.00 almost all the time ). The
second case is right answer. ;-) I suspect driver related problem.
Before FreeBSD 6.0, there were Linux Debian and I didn`t mentioned any
problem. I had there for several days FreeBSD 5.4  and watchdogs appear
too.

Vladimir