Seeking help understanding my "emX: watchdog timeout" messages

[Jack Vogel]

On the 573 get me an eeprom dump:  sysctl dev.em.4.debug=2

If you have having TX timeouts using polling, perhaps your system
is so busy that its not running the cleanup routine in time, can you
switch and run the interface in non-polled, in fact if possible using
MSI??

Jack

On Fri, Jun 20, 2008 at 1:11 PM, Rudy <crapsh at monkeybrains.net> wrote:
> Jack Vogel wrote:
>>
>> Watchdog reset is an indication that the TX completion did not happen
>> within a reasonable time, it can the symptom of a variety of problems.
>>
>> You list a bunch of settings for polling, is this interface in POLLING
>> mode?
>>
>
> Yes, DEVICE_POLLING is in the kernel and kern.polling.enable=1.  I did not
> issue a
>  "ifconfig em2 polling" and I see it running, so I assume that it is in
> POLLING mode.
>
>
> em2: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
>
>  options=1db<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,POLLING,VLAN_HWCSUM,TSO4>
>        ether 00:15:17:78:99:72
>        inet 64.215.30.154 netmask 0xfffffffc broadcast 64.215.30.155
>        media: Ethernet autoselect (1000baseTX <full-duplex>)
>        status: active
>
> Interestingly, I also get the watchdog timeouts on my em4 (motherboard
> 82573E NIC).
>
> In a 24 hour period, I see 8 'watchdog timeout' events.  4 on em2 and 4 on
> em4.  Both are uplinks to different cisco switches which vlan those ports to
> fiber uplinks.
>
> Thank you for your time Jack,
> Rudy
>